{"id":"f4d46b79-b46b-4de1-8e5d-52f28d80c88c","entityType":"agent","slug":"clawhub-cellcog-coding-agent-cellcog","name":"Coding Agent","canonicalUrl":"https://www.xpersona.co/agent/clawhub-cellcog-coding-agent-cellcog","canonicalPath":"/agent/clawhub-cellcog-coding-agent-cellcog","generatedAt":"2026-10-10T03:51:29.070Z","source":"CLAWHUB","claimStatus":"UNCLAIMED","verificationTier":"NONE","summary":{"evidence":{"source":"editorial-content","verified":true,"confidence":"high","updatedAt":"2026-10-09T15:52:35.177Z","emptyReason":null},"description":"AI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations — executed directly on your machine. Lightweight with multimedia tools loaded on demand. Skill: Coding Agent Owner: cellcog Summary: AI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations — executed directly on your machine. Lightweight with multimedia tools loaded on demand. Tags: latest:1.0.15 Version history: v1.0.15 | 2026-08-24T02:01:28.499Z | user Content updated. v1.0.14 | 2026-08-24T01:45:50.351Z | user Content updated. v1.0.","descriptionLabel":"Technical summary","evidenceSummary":"Capability contract not published. No trust telemetry is available yet. 2.4K downloads reported by the source. Last updated 10/9/2026.","installCommand":"clawhub skill install s176q1btpn094ats9b4f9hgfwx83knpk:coding-agent-cellcog","sourceUrl":"https://clawhub.ai/cellcog/coding-agent-cellcog","homepage":"https://clawhub.ai/cellcog/skills/coding-agent-cellcog","primaryLinks":[{"label":"View on ClawHub","url":"https://clawhub.ai/cellcog/coding-agent-cellcog","kind":"source"},{"label":"Homepage","url":"https://clawhub.ai/cellcog/skills/coding-agent-cellcog","kind":"homepage"}],"safetyScore":84,"overallRank":62,"popularityScore":67,"trustScore":null,"claimedByName":null,"isOwner":false,"seoDescription":"AI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations — executed directly on your machi"},"coverage":{"evidence":{"source":"public-profile","verified":false,"confidence":"medium","updatedAt":"2026-10-09T15:52:35.177Z","emptyReason":null},"protocols":[{"protocol":"OPENCLEW","label":"OpenClaw","status":"self-declared","notes":"Declared in the public agent profile."}],"capabilities":[],"verifiedCount":0,"selfDeclaredCount":1,"capabilityMatrix":{"rows":[{"key":"OPENCLEW","type":"protocol","support":"unknown","confidenceSource":"profile","notes":"Listed on profile"}],"flattenedTokens":"protocol:OPENCLEW|unknown|profile"}},"adoption":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-09T15:52:35.177Z","emptyReason":null},"stars":null,"forks":null,"downloads":2359,"packageName":null,"latestVersion":"1.0.15","tractionLabel":"2.4K downloads"},"release":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-09T15:52:35.176Z","emptyReason":null},"lastUpdatedAt":"2026-10-09T15:52:35.177Z","lastCrawledAt":"2026-10-09T15:52:35.176Z","lastIndexedAt":null,"nextCrawlAt":"2026-10-10T15:52:35.176Z","lastVerifiedAt":null,"highlights":[{"version":"1.0.15","createdAt":"2026-08-24T02:01:28.499Z","changelog":"Content updated.","fileCount":3,"zipByteSize":4968},{"version":"1.0.14","createdAt":"2026-08-24T01:45:50.351Z","changelog":"Content updated.","fileCount":3,"zipByteSize":4949},{"version":"1.0.13","createdAt":"2026-08-03T06:06:10.022Z","changelog":"Content updated.","fileCount":3,"zipByteSize":4974},{"version":"1.0.12","createdAt":"2026-08-02T20:47:09.430Z","changelog":"Display title updated.","fileCount":3,"zipByteSize":4925},{"version":"1.0.11","createdAt":"2026-07-22T15:57:21.211Z","changelog":"- The skill name was updated from \"code-cog\" to \"coding-agent-cellcog\" for consistency and clarity. - Minor edits to documentation and section headers to improve clarity. - Removed the redundant file \"skill-card.md\" from the skill package. - Updated internal references to align with the new skill name. - No changes to core functionality or APIs.","fileCount":3,"zipByteSize":4888},{"version":"1.0.10","createdAt":"2026-04-23T06:18:36.013Z","changelog":"- Added explicit Windows OS support in metadata. - Declared requirements for `python3` binary and `CELLCOG_API_KEY` environment variable in skill metadata. - No functional or API changes; documentation and compatibility metadata only.","fileCount":3,"zipByteSize":4962},{"version":"1.0.9","createdAt":"2026-04-14T17:31:49.013Z","changelog":"- Updated skill and description for clarity; emphasized AI agent use and CellCog Co-work integration. - Clarified SDK usage and agent provider distinctions, especially for OpenClaw vs. other agents. - Streamlined language throughout for conciseness and better readability. - Improved prerequisites and setup instructions for easier onboarding. - No functional changes; documentation improvements only.","fileCount":2,"zipByteSize":3699},{"version":"1.0.8","createdAt":"2026-04-13T01:00:25.534Z","changelog":"- Refined the skill description for clarity and conciseness. - Updated usage instructions for better agent compatibility (notably Cursor, Claude Code, etc.). - Improved quick start and code examples for consistency. - Enhanced general documentation layout and omitted some redundant content. - No changes to code or core functionality.","fileCount":2,"zipByteSize":3711}]},"execution":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No published capability contract is available yet."},"installCommand":"clawhub skill install s176q1btpn094ats9b4f9hgfwx83knpk:coding-agent-cellcog","setupComplexity":"low","setupSteps":["Setup complexity is classified as HIGH. You must provision dedicated cloud infrastructure or an isolated VM. Do not run this directly on your local workstation.","Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data."],"contract":{"contractStatus":"missing","authModes":[],"requires":[],"forbidden":[],"supportsMcp":false,"supportsA2a":false,"supportsStreaming":false,"inputSchemaRef":null,"outputSchemaRef":null,"dataRegion":null,"contractUpdatedAt":null,"sourceUpdatedAt":null,"freshnessSeconds":null},"invocationGuide":{"preferredApi":{"snapshotUrl":"https://www.xpersona.co/api/v1/agents/clawhub-cellcog-coding-agent-cellcog/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-cellcog-coding-agent-cellcog/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-cellcog-coding-agent-cellcog/trust"},"curlExamples":["curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-cellcog-coding-agent-cellcog/snapshot\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-cellcog-coding-agent-cellcog/contract\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-cellcog-coding-agent-cellcog/trust\""],"jsonRequestTemplate":{"query":"summarize this repo","constraints":{"maxLatencyMs":2000,"protocolPreference":["OPENCLEW"]}},"jsonResponseTemplate":{"ok":true,"result":{"summary":"...","confidence":0.9},"meta":{"source":"CLAWHUB","generatedAt":"2026-10-10T03:51:29.066Z"}},"retryPolicy":{"maxAttempts":3,"backoffMs":[500,1500,3500],"retryableConditions":["HTTP_429","HTTP_503","NETWORK_TIMEOUT"]}},"endpoints":{"dossierUrl":"https://www.xpersona.co/api/v1/agents/clawhub-cellcog-coding-agent-cellcog/dossier","snapshotUrl":"https://www.xpersona.co/api/v1/agents/clawhub-cellcog-coding-agent-cellcog/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-cellcog-coding-agent-cellcog/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-cellcog-coding-agent-cellcog/trust"}},"reliability":{"evidence":{"source":"runtime-metrics","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No trust, reliability, or runtime telemetry is available."},"trust":{"status":"unavailable","handshakeStatus":"UNKNOWN","verificationFreshnessHours":null,"reputationScore":null,"p95LatencyMs":null,"successRate30d":null,"fallbackRate":null,"attempts30d":null,"trustUpdatedAt":null,"trustConfidence":"unknown","sourceUpdatedAt":null,"freshnessSeconds":null},"decisionGuardrails":{"doNotUseIf":["Contract metadata is missing or unavailable for deterministic execution."],"safeUseWhen":[],"riskFlags":["missing_or_unavailable_contract","trust_data_unavailable","schema_references_missing"],"operationalConfidence":"low"},"executionMetrics":{"observedLatencyMsP50":null,"observedLatencyMsP95":null,"estimatedCostUsd":null,"uptime30d":null,"rateLimitRpm":null,"rateLimitBurst":null,"lastVerifiedAt":null,"verificationSource":null},"runtimeMetrics":{"successRate":null,"avgLatencyMs":null,"avgCostUsd":null,"hallucinationRate":null,"retryRate":null,"disputeRate":null,"p50Latency":null,"p95Latency":null,"lastUpdated":null}},"benchmarks":{"evidence":{"source":"no-benchmark-data","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No benchmark suites or observed failure patterns are available."},"suites":[],"failurePatterns":[]},"artifacts":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"high","updatedAt":"2026-10-09T15:52:35.177Z","emptyReason":null},"readme":"Skill: Coding Agent\n\nOwner: cellcog\n\nSummary: AI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations — executed directly on your machine. Lightweight with multimedia tools loaded on demand.\n\nTags: latest:1.0.15\n\nVersion history:\n\nv1.0.15 | 2026-08-24T02:01:28.499Z | user\n\nContent updated.\n\nv1.0.14 | 2026-08-24T01:45:50.351Z | user\n\nContent updated.\n\nv1.0.13 | 2026-08-03T06:06:10.022Z | user\n\nContent updated.\n\nv1.0.12 | 2026-08-02T20:47:09.430Z | user\n\nDisplay title updated.\n\nv1.0.11 | 2026-07-22T15:57:21.211Z | auto\n\n- The skill name was updated from \"code-cog\" to \"coding-agent-cellcog\" for consistency and clarity.\n- Minor edits to documentation and section headers to improve clarity.\n- Removed the redundant file \"skill-card.md\" from the skill package.\n- Updated internal references to align with the new skill name.\n- No changes to core functionality or APIs.\n\nv1.0.10 | 2026-04-23T06:18:36.013Z | auto\n\n- Added explicit Windows OS support in metadata.\n- Declared requirements for `python3` binary and `CELLCOG_API_KEY` environment variable in skill metadata.\n- No functional or API changes; documentation and compatibility metadata only.\n\nv1.0.9 | 2026-04-14T17:31:49.013Z | auto\n\n- Updated skill and description for clarity; emphasized AI agent use and CellCog Co-work integration.\n- Clarified SDK usage and agent provider distinctions, especially for OpenClaw vs. other agents.\n- Streamlined language throughout for conciseness and better readability.\n- Improved prerequisites and setup instructions for easier onboarding.\n- No functional changes; documentation improvements only.\n\nv1.0.8 | 2026-04-13T01:00:25.534Z | auto\n\n- Refined the skill description for clarity and conciseness.\n- Updated usage instructions for better agent compatibility (notably Cursor, Claude Code, etc.).\n- Improved quick start and code examples for consistency.\n- Enhanced general documentation layout and omitted some redundant content.\n- No changes to code or core functionality.\n\nv1.0.7 | 2026-04-12T23:35:01.360Z | auto\n\n- SKILL.md has been streamlined with a more concise description and a shorter \"How to Use\" section at the top.\n- References to the cellcog SDK documentation and homepage have been updated for clarity.\n- The overall documentation has been trimmed for brevity, with redundant and explanatory content reduced or omitted.\n- Quickstart examples and prerequisites remain, but non-essential repetition and extra details have been removed.\n- No code or feature changes—this update is documentation only.\n\nv1.0.6 | 2026-04-11T05:52:28.003Z | auto\n\n- Updated SDK usage examples: replaced agent_name with agent_provider in all code snippets.\n- No functional or feature changes; documentation update only.\n\nv1.0.5 | 2026-04-11T02:58:39.914Z | auto\n\n- Added agent_name=\"openclaw\" to all CellCogClient code examples for clarity on agent initialization.\n- No changes to logic, features, or behavior; documentation improvements only.\n\nv1.0.4 | 2026-04-08T05:57:37.117Z | auto\n\n- Major SKILL.md rewrite for clarity, conciseness, and usability.\n- Streamlined setup instructions and moved Co-work requirements to a dedicated section.\n- Added new \"Quick Start,\" \"Tips for Better Results,\" and expanded usage examples with parameter explanations.\n- Clearly outlined capabilities: code generation, debugging, refactoring, terminal operations, and codebase exploration.\n- Emphasized agent-first design, direct machine access via Co-work, and safety/approval model.\n- Noted current limitations (macOS/Linux only, user approval for writes).\n\nv1.0.3 | 2026-04-06T05:06:50.683Z | auto\n\n- Improved and clarified skill description to emphasize direct machine access and suitability for agent workflows.\n- Updated prerequisites and quick start guides with references to the Cowork-Cog skill for CellCog Desktop setup.\n- Simplified and reorganized documentation for easier readability, focusing on core capabilities, requirements, and workflow.\n- Consolidated explanation of features, removing repetitive detail and focusing on key functions such as code generation, debugging, and terminal operations.\n- Added a \"Related Skills\" section to cross-link relevant CellCog skills.\n- No code or interface changes; documentation update only.\n\nv1.0.2 | 2026-04-03T01:43:04.141Z | auto\n\n- Added usage instructions for OpenClaw agents, including the notify_session_key parameter for fire-and-forget workflows.\n- Clarified the difference between OpenClaw and other agent usage in the Quick Start section.\n- No functional or API changes; documentation only.\n\nv1.0.1 | 2026-04-03T00:07:28.102Z | auto\n\n- Updated example code and parameter names to match the current SDK (`enable_cowork`, `cowork_working_directory`, `task_label`)\n- Quick Start example now blocks until results are returned, clarifying the delivery model\n- Refreshed and streamlined \"Example Prompts\" for clarity and real-world relevance\n- Added a reference to consult the **cellcog** mothership skill for full SDK/API usage\n- Deprecated parameters (`hc_enabled`, `hc_working_directory`, `notify_session_key`) have been removed from docs and examples\n\nv1.0.0 | 2026-03-31T20:12:18.600Z | auto\n\n- Initial release of CodeCog: a coding agent built for AI agents, enabling direct code generation, debugging, refactoring, and terminal operations on the user's machine via CellCog Co-work.\n- Supports lightweight operation with optional on-demand multimedia tool loading.\n- Integrates with CellCog Desktop for secure file system and terminal access (macOS/Linux only).\n- Provides examples for code generation, bug fixing, refactoring, and test creation.\n- Ensures user safety with approval workflows for writes/executes and auto-approve options.\n- Reads project conventions from AGENTS.md and adapts actions to existing codebase patterns.\n\nArchive index:\n\nArchive v1.0.15: 3 files, 4968 bytes\n\nFiles: skill-card.md (2145b), SKILL.md (9270b), _meta.json (140b)\n\nFile v1.0.15:SKILL.md\n\n---\nname: coding-agent-cellcog\ndescription: \"AI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations — executed directly on your machine. Lightweight with multimedia tools loaded on demand.\"\nauthor: CellCog\nhomepage: https://cellcog.ai\nmetadata:\n  openclaw:\n    emoji: \"💻\"\n    os: [darwin, linux, windows]\n    requires:\n      bins: [python3]\n      env: [CELLCOG_API_KEY]\ndependencies: [cellcog]\n---\n# Coding Agent — The First Coding Agent Built for Agents\n\nWhen your AI needs to code, it delegates to CodeCog. Direct codebase access, terminal operations, and file editing — executed on the user's machine via CellCog Co-work.\n\n## How to Use\n\nFor your first CellCog task in a session, read the **cellcog** skill for the full SDK reference — file handling, chat modes, timeouts, and more.\n\n**OpenClaw (fire-and-forget):**\n```python\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    notify_session_key=\"agent:main:main\",\n    task_label=\"my-task\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\n```\n\n**All agents except OpenClaw (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw|cursor|claude-code|codex|...\")\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    task_label=\"my-task\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\nprint(result[\"message\"])\n```\n\n\n---\n\n## Prerequisites\n\nThis skill requires the `cellcog` skill for SDK setup and API calls.\n\n```bash\n# Claude Code, Cursor, Codex + 70 more agents\nnpx skills add cellcog/skills --skill cellcog\n\n# OpenClaw\nopenclaw skills install @cellcog/cellcog\n```\n\n**Read the cellcog skill first** for SDK setup. This skill shows you how to use CellCog as a coding agent.\n\n**CellCog Desktop Required:** The user must have CellCog Desktop installed and running for Co-work (direct machine access). Download at https://cellcog.ai\n\n---\n\n## Quick Start\n\n**OpenClaw agents (fire-and-forget):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    notify_session_key=\"agent:main:main\",  # OpenClaw only\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**All other agents (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**Key parameters:**\n- `chat_mode=\"agent\", chat_tier=\"max\"` — coding needs the deepest reasoning tier (the SDK applies `\"max\"` automatically when `enable_cowork=True`)\n- `enable_cowork=True` — Enables Co-work (direct machine access)\n- `cowork_working_directory` — The repo/directory to work in\n\n---\n\n## What CodeCog Can Do\n\n### Code Generation & Editing\n- Write new files, modules, and components\n- Edit existing code with surgical precision\n- Refactor codebases — rename, restructure, extract\n- Port code between languages or frameworks\n\n### Debugging & Fixing\n- Read error logs and stack traces\n- Identify root causes across multiple files\n- Apply fixes and verify they work\n- Run tests to confirm the fix\n\n### Terminal Operations\n- Run build commands, tests, linters\n- Install dependencies (npm, pip, cargo, etc.)\n- Git operations (status, diff, commit)\n- Docker, deployment scripts\n\n### Codebase Exploration\n- Auto-reads AGENTS.md/CLAUDE.md for project conventions\n- Explores directory structure before starting work\n- Understands existing patterns and follows them\n- Reads related files to maintain consistency\n\n---\n\n## What Makes CodeCog Different\n\n### Built for Agents, Not Humans\n\nEvery other coding tool (Cursor, Claude Code, Codex, Windsurf) is designed for human developers sitting at a screen. CodeCog is designed for AI agents that need to code programmatically — fire a request, get results back, continue orchestrating.\n\n### Starts Lean, Scales to Multimodal\n\nCodeCog runs CellCog's agent mode at the max tier with a lean, coding-focused context. But if your task unexpectedly needs images, PDFs, videos, or other capabilities, the agent loads those tools on demand. No other coding agent does this.\n\nExample: Your agent asks CodeCog to set up a new project. CodeCog writes the code, then realizes it needs to generate a logo for the README — it loads image tools, generates the logo, and continues. Seamless.\n\n### Direct Machine Access\n\nVia CellCog Co-work, CodeCog operates directly on the user's filesystem:\n- Reads and writes files on the real machine\n- Runs terminal commands in the user's shell\n- Respects project conventions (AGENTS.md, .gitignore, etc.)\n- User approves write/execute operations for safety\n\n---\n\n## Choosing Mode & Tier\n\n**Use `chat_mode=\"agent\", chat_tier=\"max\"` for all coding work** — code needs the deepest reasoning tier. The SDK applies `\"max\"` automatically whenever `enable_cowork=True`, so co-work sessions get it even without an explicit tier.\n\n`\"agent core\"` is a legacy name that still works forever (the server maps it to Agent max), but new code should pass `chat_mode=\"agent\", chat_tier=\"max\"`.\n\nAgent Team (`chat_mode=\"team\"`) is reserved for deep research — use it only when the task IS research that happens to involve code.\n\n---\n\n## Example Prompts\n\n### New Feature Development\n\n```python\nresult = client.create_chat(\n    prompt=\"Add a REST API endpoint for user profile updates with validation and tests\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"add-profile-api\",\n)\n```\n\n### Bug Fix from Error Log\n\n```python\nresult = client.create_chat(\n    prompt=\"\"\"Fix this error in production:\nTypeError: Cannot read properties of undefined (reading 'map')\nat UserList.render (src/components/UserList.tsx:42)\n\nThe component crashes when the API returns an empty response.\"\"\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"fix-userlist-crash\",\n)\n```\n\n### Codebase Refactor\n\n```python\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module from session-based to JWT tokens. Update all middleware, tests, and API routes.\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n### Test Generation\n\n```python\nresult = client.create_chat(\n    prompt=\"Generate comprehensive unit tests for src/services/billing.py. Cover edge cases for proration, currency conversion, and failed payments.\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"billing-tests\",\n)\n```\n\nSee https://cellcog.ai for complete SDK API reference — delivery modes, `send_message()`, timeouts, file handling, and more.\n\n---\n\n## Co-work Setup\n\n### Requirements\n1. **CellCog Desktop** must be installed and running on the user's machine\n2. **Working directory** must be specified — this is the root of the project/repo\n3. User must be logged into CellCog Desktop with the same account\n\n### What Co-work Enables\n- `HumanComputer_Terminal` — Run shell commands on the user's machine\n- `HumanComputer_Terminal_File_View` — Read files on the user's machine\n- `HumanComputer_Terminal_File_Write` — Write files on the user's machine\n- `HumanComputer_Terminal_File_Edit` — Edit files on the user's machine\n\n### Safety Model\n- **Read operations** are auto-approved (no interruption)\n- **Write/execute operations** require user approval in the CellCog web UI\n- Users can configure auto-approve for reads/writes within the working directory\n- Sensitive paths (credentials, SSH keys) are always blocked\n\n---\n\n## Tips for Better Results\n\n1. **Specify the working directory** — Always set `cowork_working_directory` to the project root\n2. **Reference specific files** — \"Fix the bug in src/auth/login.ts\" is better than \"fix the login bug\"\n3. **Mention conventions** — \"Follow the existing test patterns\" helps maintain consistency\n4. **Include error context** — Stack traces, log output, and reproduction steps help debugging\n5. **Use AGENTS.md** — Place an AGENTS.md at your repo root with build commands, style guides, and project structure. CodeCog reads it automatically.\n\n---\n\n## Limitations\n\n- **macOS and Linux only** — CellCog Desktop (Co-work) is not yet available on Windows\n- **CellCog Desktop required** — Without Co-work, CodeCog can still write code in its Docker workspace, but cannot access the user's machine directly\n- **User approval for writes** — Write operations pause for user approval (configurable auto-approve available)\n\nFile v1.0.15:_meta.json\n\n{\n  \"ownerId\": \"kn7a96cj9q65e0bhmzahv790en80ffqm\",\n  \"slug\": \"coding-agent-cellcog\",\n  \"version\": \"1.0.15\",\n  \"publishedAt\": 1787536888499\n}\n\nFile v1.0.15:skill-card.md\n\n## Description:\n\nAI coding agent powered by CellCog Co-work for code generation, debugging, refactoring, codebase exploration, and terminal operations executed on the user's machine.\n\nThis skill is ready for commercial/non-commercial use.\n\n## Publisher:\n\n[cellcog](https://clawhub.ai/user/cellcog)\n\n### License/Terms of Use:\n\nMIT-0\n\n## Use Case:\n\nDevelopers and agents use this skill to delegate coding work to CellCog Co-work, including code generation, bug fixing, refactoring, repository exploration, tests, and terminal operations in a specified project directory.\n\n### Deployment Geography for Use:\n\nGlobal\n\n## Known Risks and Mitigations:\n\nRisk: CellCog Co-work gives a coding agent local project access for file reads, file writes, and terminal commands.\n\nMitigation: Scope Co-work to a single project directory and review write and command actions before approval.\n\nRisk: Configurable write auto-approval can allow broader changes inside the selected working directory.\n\nMitigation: Leave write auto-approve disabled unless the installed CellCog components and project scope have been verified.\n\nRisk: The skill depends on separately installed CellCog Desktop/runtime components.\n\nMitigation: Install only if you trust CellCog and have verified the installed CellCog versions.\n\n## Reference(s):\n\n- [CellCog](https://cellcog.ai)\n- [ClawHub Coding Agent Skill](https://clawhub.ai/cellcog/skills/coding-agent-cellcog)\n\n## Skill Output:\n\n**Output Type(s):** [Text, Markdown, Code, Shell commands, Configuration, Guidance]\n\n**Output Format:** [Markdown and text with inline code, shell commands, and configuration snippets]\n\n**Output Parameters:** [1D]\n\n**Other Properties Related to Output:** [May produce file edits and terminal actions through CellCog Co-work subject to the user's local approval settings.]\n\n## Skill Version(s):\n\n1.0.15 (source: server release metadata)\n\n## Ethical Considerations:\n\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment.\n\nArchive v1.0.14: 3 files, 4949 bytes\n\nFiles: skill-card.md (2122b), SKILL.md (9253b), _meta.json (140b)\n\nFile v1.0.14:SKILL.md\n\n---\nname: coding-agent-cellcog\ndescription: \"AI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations — executed directly on your machine. Lightweight with multimedia tools loaded on demand.\"\nauthor: CellCog\nhomepage: https://cellcog.ai\nmetadata:\n  openclaw:\n    emoji: \"💻\"\n    os: [darwin, linux, windows]\n    requires:\n      bins: [python3]\n      env: [CELLCOG_API_KEY]\ndependencies: [cellcog]\n---\n# Coding Agent — The First Coding Agent Built for Agents\n\nWhen your AI needs to code, it delegates to CodeCog. Direct codebase access, terminal operations, and file editing — executed on the user's machine via CellCog Co-work.\n\n## How to Use\n\nFor your first CellCog task in a session, read the **cellcog** skill for the full SDK reference — file handling, chat modes, timeouts, and more.\n\n**OpenClaw (fire-and-forget):**\n```python\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    notify_session_key=\"agent:main:main\",\n    task_label=\"my-task\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\n```\n\n**All agents except OpenClaw (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw|cursor|claude-code|codex|...\")\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    task_label=\"my-task\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\nprint(result[\"message\"])\n```\n\n\n---\n\n## Prerequisites\n\nThis skill requires the `cellcog` skill for SDK setup and API calls.\n\n```bash\n# Claude Code, Cursor, Codex + 70 more agents\nnpx skills add cellcog/skills --skill cellcog\n\n# OpenClaw\nclawhub install cellcog\n```\n\n**Read the cellcog skill first** for SDK setup. This skill shows you how to use CellCog as a coding agent.\n\n**CellCog Desktop Required:** The user must have CellCog Desktop installed and running for Co-work (direct machine access). Download at https://cellcog.ai\n\n---\n\n## Quick Start\n\n**OpenClaw agents (fire-and-forget):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    notify_session_key=\"agent:main:main\",  # OpenClaw only\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**All other agents (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**Key parameters:**\n- `chat_mode=\"agent\", chat_tier=\"max\"` — coding needs the deepest reasoning tier (the SDK applies `\"max\"` automatically when `enable_cowork=True`)\n- `enable_cowork=True` — Enables Co-work (direct machine access)\n- `cowork_working_directory` — The repo/directory to work in\n\n---\n\n## What CodeCog Can Do\n\n### Code Generation & Editing\n- Write new files, modules, and components\n- Edit existing code with surgical precision\n- Refactor codebases — rename, restructure, extract\n- Port code between languages or frameworks\n\n### Debugging & Fixing\n- Read error logs and stack traces\n- Identify root causes across multiple files\n- Apply fixes and verify they work\n- Run tests to confirm the fix\n\n### Terminal Operations\n- Run build commands, tests, linters\n- Install dependencies (npm, pip, cargo, etc.)\n- Git operations (status, diff, commit)\n- Docker, deployment scripts\n\n### Codebase Exploration\n- Auto-reads AGENTS.md/CLAUDE.md for project conventions\n- Explores directory structure before starting work\n- Understands existing patterns and follows them\n- Reads related files to maintain consistency\n\n---\n\n## What Makes CodeCog Different\n\n### Built for Agents, Not Humans\n\nEvery other coding tool (Cursor, Claude Code, Codex, Windsurf) is designed for human developers sitting at a screen. CodeCog is designed for AI agents that need to code programmatically — fire a request, get results back, continue orchestrating.\n\n### Starts Lean, Scales to Multimodal\n\nCodeCog runs CellCog's agent mode at the max tier with a lean, coding-focused context. But if your task unexpectedly needs images, PDFs, videos, or other capabilities, the agent loads those tools on demand. No other coding agent does this.\n\nExample: Your agent asks CodeCog to set up a new project. CodeCog writes the code, then realizes it needs to generate a logo for the README — it loads image tools, generates the logo, and continues. Seamless.\n\n### Direct Machine Access\n\nVia CellCog Co-work, CodeCog operates directly on the user's filesystem:\n- Reads and writes files on the real machine\n- Runs terminal commands in the user's shell\n- Respects project conventions (AGENTS.md, .gitignore, etc.)\n- User approves write/execute operations for safety\n\n---\n\n## Choosing Mode & Tier\n\n**Use `chat_mode=\"agent\", chat_tier=\"max\"` for all coding work** — code needs the deepest reasoning tier. The SDK applies `\"max\"` automatically whenever `enable_cowork=True`, so co-work sessions get it even without an explicit tier.\n\n`\"agent core\"` is a legacy name that still works forever (the server maps it to Agent max), but new code should pass `chat_mode=\"agent\", chat_tier=\"max\"`.\n\nAgent Team (`chat_mode=\"team\"`) is reserved for deep research — use it only when the task IS research that happens to involve code.\n\n---\n\n## Example Prompts\n\n### New Feature Development\n\n```python\nresult = client.create_chat(\n    prompt=\"Add a REST API endpoint for user profile updates with validation and tests\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"add-profile-api\",\n)\n```\n\n### Bug Fix from Error Log\n\n```python\nresult = client.create_chat(\n    prompt=\"\"\"Fix this error in production:\nTypeError: Cannot read properties of undefined (reading 'map')\nat UserList.render (src/components/UserList.tsx:42)\n\nThe component crashes when the API returns an empty response.\"\"\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"fix-userlist-crash\",\n)\n```\n\n### Codebase Refactor\n\n```python\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module from session-based to JWT tokens. Update all middleware, tests, and API routes.\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n### Test Generation\n\n```python\nresult = client.create_chat(\n    prompt=\"Generate comprehensive unit tests for src/services/billing.py. Cover edge cases for proration, currency conversion, and failed payments.\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"billing-tests\",\n)\n```\n\nSee https://cellcog.ai for complete SDK API reference — delivery modes, `send_message()`, timeouts, file handling, and more.\n\n---\n\n## Co-work Setup\n\n### Requirements\n1. **CellCog Desktop** must be installed and running on the user's machine\n2. **Working directory** must be specified — this is the root of the project/repo\n3. User must be logged into CellCog Desktop with the same account\n\n### What Co-work Enables\n- `HumanComputer_Terminal` — Run shell commands on the user's machine\n- `HumanComputer_Terminal_File_View` — Read files on the user's machine\n- `HumanComputer_Terminal_File_Write` — Write files on the user's machine\n- `HumanComputer_Terminal_File_Edit` — Edit files on the user's machine\n\n### Safety Model\n- **Read operations** are auto-approved (no interruption)\n- **Write/execute operations** require user approval in the CellCog web UI\n- Users can configure auto-approve for reads/writes within the working directory\n- Sensitive paths (credentials, SSH keys) are always blocked\n\n---\n\n## Tips for Better Results\n\n1. **Specify the working directory** — Always set `cowork_working_directory` to the project root\n2. **Reference specific files** — \"Fix the bug in src/auth/login.ts\" is better than \"fix the login bug\"\n3. **Mention conventions** — \"Follow the existing test patterns\" helps maintain consistency\n4. **Include error context** — Stack traces, log output, and reproduction steps help debugging\n5. **Use AGENTS.md** — Place an AGENTS.md at your repo root with build commands, style guides, and project structure. CodeCog reads it automatically.\n\n---\n\n## Limitations\n\n- **macOS and Linux only** — CellCog Desktop (Co-work) is not yet available on Windows\n- **CellCog Desktop required** — Without Co-work, CodeCog can still write code in its Docker workspace, but cannot access the user's machine directly\n- **User approval for writes** — Write operations pause for user approval (configurable auto-approve available)\n\nFile v1.0.14:_meta.json\n\n{\n  \"ownerId\": \"kn7a96cj9q65e0bhmzahv790en80ffqm\",\n  \"slug\": \"coding-agent-cellcog\",\n  \"version\": \"1.0.14\",\n  \"publishedAt\": 1787535950351\n}\n\nFile v1.0.14:skill-card.md\n\n## Description:\n\nAI coding agent powered by CellCog Co-work for code generation, debugging, refactoring, codebase exploration, and terminal operations executed on the user's machine.\n\nThis skill is ready for commercial/non-commercial use.\n\n## Publisher:\n\n[cellcog](https://clawhub.ai/user/cellcog)\n\n### License/Terms of Use:\n\nMIT-0\n\n## Use Case:\n\nDevelopers and agent builders use this skill to delegate coding work to CellCog Co-work, including code generation, debugging, refactoring, codebase exploration, and terminal-based verification in a selected local project.\n\n### Deployment Geography for Use:\n\nGlobal\n\n## Known Risks and Mitigations:\n\nRisk: The skill can enable coding work against files in a selected local project and can involve terminal operations.\n\nMitigation: Use the smallest appropriate cowork_working_directory, keep write and execute approvals manual unless the repository and task are trusted, and review proposed changes before accepting them.\n\nRisk: Auto-approved reads may expose project contents within the selected working directory.\n\nMitigation: Avoid using the skill on sensitive repositories unless local read access is acceptable for the task.\n\n## Reference(s):\n\n- [ClawHub Skill Page](https://clawhub.ai/cellcog/skills/coding-agent-cellcog)\n- [CellCog](https://cellcog.ai)\n- [CellCog Publisher Profile](https://clawhub.ai/user/cellcog)\n\n## Skill Output:\n\n**Output Type(s):** [Text, Markdown, Code, Shell commands, Configuration, Guidance]\n\n**Output Format:** [Markdown or text responses with code blocks, shell commands, and configuration snippets as needed]\n\n**Output Parameters:** [1D]\n\n**Other Properties Related to Output:** [May propose or perform local file and terminal operations through CellCog Co-work, subject to the user's configured approvals.]\n\n## Skill Version(s):\n\n1.0.14 (source: server release evidence)\n\n## Ethical Considerations:\n\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment.\n\nArchive v1.0.13: 3 files, 4974 bytes\n\nFiles: skill-card.md (2267b), SKILL.md (9012b), _meta.json (140b)\n\nFile v1.0.13:SKILL.md\n\n---\nname: coding-agent-cellcog\ndescription: \"AI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations — executed directly on your machine. Lightweight with multimedia tools loaded on demand.\"\nauthor: CellCog\nhomepage: https://cellcog.ai\nmetadata:\n  openclaw:\n    emoji: \"💻\"\n    os: [darwin, linux, windows]\n    requires:\n      bins: [python3]\n      env: [CELLCOG_API_KEY]\ndependencies: [cellcog]\n---\n# Coding Agent — The First Coding Agent Built for Agents\n\nWhen your AI needs to code, it delegates to CodeCog. Direct codebase access, terminal operations, and file editing — executed on the user's machine via CellCog Co-work.\n\n## How to Use\n\nFor your first CellCog task in a session, read the **cellcog** skill for the full SDK reference — file handling, chat modes, timeouts, and more.\n\n**OpenClaw (fire-and-forget):**\n```python\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    notify_session_key=\"agent:main:main\",\n    task_label=\"my-task\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\n```\n\n**All agents except OpenClaw (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw|cursor|claude-code|codex|...\")\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    task_label=\"my-task\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\nprint(result[\"message\"])\n```\n\n\n---\n\n## Prerequisites\n\nThis skill requires the `cellcog` skill for SDK setup and API calls.\n\n```bash\n# Claude Code, Cursor, Codex + 70 more agents\nnpx skills add cellcog/skills --skill cellcog\n\n# OpenClaw\nclawhub install cellcog\n```\n\n**Read the cellcog skill first** for SDK setup. This skill shows you how to use CellCog as a coding agent.\n\n**CellCog Desktop Required:** The user must have CellCog Desktop installed and running for Co-work (direct machine access). Download at https://cellcog.ai\n\n---\n\n## Quick Start\n\n**OpenClaw agents (fire-and-forget):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    notify_session_key=\"agent:main:main\",  # OpenClaw only\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**All other agents (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**Key parameters:**\n- `chat_mode=\"agent core\"` — Lightweight coding agent (vs `\"agent\"` for full multimedia)\n- `enable_cowork=True` — Enables Co-work (direct machine access)\n- `cowork_working_directory` — The repo/directory to work in\n\n---\n\n## What CodeCog Can Do\n\n### Code Generation & Editing\n- Write new files, modules, and components\n- Edit existing code with surgical precision\n- Refactor codebases — rename, restructure, extract\n- Port code between languages or frameworks\n\n### Debugging & Fixing\n- Read error logs and stack traces\n- Identify root causes across multiple files\n- Apply fixes and verify they work\n- Run tests to confirm the fix\n\n### Terminal Operations\n- Run build commands, tests, linters\n- Install dependencies (npm, pip, cargo, etc.)\n- Git operations (status, diff, commit)\n- Docker, deployment scripts\n\n### Codebase Exploration\n- Auto-reads AGENTS.md/CLAUDE.md for project conventions\n- Explores directory structure before starting work\n- Understands existing patterns and follows them\n- Reads related files to maintain consistency\n\n---\n\n## What Makes CodeCog Different\n\n### Built for Agents, Not Humans\n\nEvery other coding tool (Cursor, Claude Code, Codex, Windsurf) is designed for human developers sitting at a screen. CodeCog is designed for AI agents that need to code programmatically — fire a request, get results back, continue orchestrating.\n\n### Starts Lean, Scales to Multimodal\n\nCodeCog uses CellCog's Agent Core mode — a lightweight context focused on coding. But if your task unexpectedly needs images, PDFs, videos, or other capabilities, the agent loads those tools on demand. No other coding agent does this.\n\nExample: Your agent asks CodeCog to set up a new project. CodeCog writes the code, then realizes it needs to generate a logo for the README — it loads image tools, generates the logo, and continues. Seamless.\n\n### Direct Machine Access\n\nVia CellCog Co-work, CodeCog operates directly on the user's filesystem:\n- Reads and writes files on the real machine\n- Runs terminal commands in the user's shell\n- Respects project conventions (AGENTS.md, .gitignore, etc.)\n- User approves write/execute operations for safety\n\n---\n\n## Chat Mode\n\n**Always use `\"agent core\"` for CodeCog.** This is the dedicated lightweight mode optimized for coding.\n\n| Mode | Use Case |\n|------|----------|\n| `\"agent core\"` | **CodeCog default** — coding, co-work, terminal ops (50 credits min) |\n| `\"agent\"` | Full multimedia agent — use when you need images/video/audio alongside code (100 credits min) |\n| `\"agent team\"` | Deep research + coding — use for architecture decisions or complex refactors needing research (500 credits min) |\n\n---\n\n## Example Prompts\n\n### New Feature Development\n\n```python\nresult = client.create_chat(\n    prompt=\"Add a REST API endpoint for user profile updates with validation and tests\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"add-profile-api\",\n)\n```\n\n### Bug Fix from Error Log\n\n```python\nresult = client.create_chat(\n    prompt=\"\"\"Fix this error in production:\nTypeError: Cannot read properties of undefined (reading 'map')\nat UserList.render (src/components/UserList.tsx:42)\n\nThe component crashes when the API returns an empty response.\"\"\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"fix-userlist-crash\",\n)\n```\n\n### Codebase Refactor\n\n```python\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module from session-based to JWT tokens. Update all middleware, tests, and API routes.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n### Test Generation\n\n```python\nresult = client.create_chat(\n    prompt=\"Generate comprehensive unit tests for src/services/billing.py. Cover edge cases for proration, currency conversion, and failed payments.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"billing-tests\",\n)\n```\n\nSee https://cellcog.ai for complete SDK API reference — delivery modes, `send_message()`, timeouts, file handling, and more.\n\n---\n\n## Co-work Setup\n\n### Requirements\n1. **CellCog Desktop** must be installed and running on the user's machine\n2. **Working directory** must be specified — this is the root of the project/repo\n3. User must be logged into CellCog Desktop with the same account\n\n### What Co-work Enables\n- `HumanComputer_Terminal` — Run shell commands on the user's machine\n- `HumanComputer_Terminal_File_View` — Read files on the user's machine\n- `HumanComputer_Terminal_File_Write` — Write files on the user's machine\n- `HumanComputer_Terminal_File_Edit` — Edit files on the user's machine\n\n### Safety Model\n- **Read operations** are auto-approved (no interruption)\n- **Write/execute operations** require user approval in the CellCog web UI\n- Users can configure auto-approve for reads/writes within the working directory\n- Sensitive paths (credentials, SSH keys) are always blocked\n\n---\n\n## Tips for Better Results\n\n1. **Specify the working directory** — Always set `cowork_working_directory` to the project root\n2. **Reference specific files** — \"Fix the bug in src/auth/login.ts\" is better than \"fix the login bug\"\n3. **Mention conventions** — \"Follow the existing test patterns\" helps maintain consistency\n4. **Include error context** — Stack traces, log output, and reproduction steps help debugging\n5. **Use AGENTS.md** — Place an AGENTS.md at your repo root with build commands, style guides, and project structure. CodeCog reads it automatically.\n\n---\n\n## Limitations\n\n- **macOS and Linux only** — CellCog Desktop (Co-work) is not yet available on Windows\n- **CellCog Desktop required** — Without Co-work, CodeCog can still write code in its Docker workspace, but cannot access the user's machine directly\n- **User approval for writes** — Write operations pause for user approval (configurable auto-approve available)\n\nFile v1.0.13:_meta.json\n\n{\n  \"ownerId\": \"kn7a96cj9q65e0bhmzahv790en80ffqm\",\n  \"slug\": \"coding-agent-cellcog\",\n  \"version\": \"1.0.13\",\n  \"publishedAt\": 1785737170022\n}\n\nFile v1.0.13:skill-card.md\n\n## Description: <br>\nAI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations - executed directly on your machine. Lightweight with multimedia tools loaded on demand. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[cellcog](https://clawhub.ai/user/cellcog) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers and AI agents use this skill to delegate coding tasks to CodeCog through CellCog Co-work, including code generation, debugging, refactoring, codebase exploration, and terminal operations in a selected project directory. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: The skill can let an AI coding agent read, edit, and run commands in a selected project directory through CellCog Desktop. <br>\nMitigation: Install only when that project-level access is acceptable, choose the working directory deliberately, and keep write auto-approval disabled unless the repository and task are trusted. <br>\nRisk: Terminal, dependency, git, Docker, or deployment actions can change the local project or environment. <br>\nMitigation: Review requested terminal and write actions before approval, especially commands that install dependencies, modify version control state, or deploy software. <br>\n\n\n## Reference(s): <br>\n- [CellCog](https://cellcog.ai) <br>\n- [ClawHub Skill Page](https://clawhub.ai/cellcog/skills/coding-agent-cellcog) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [Text, Markdown, Code, Shell commands, Configuration, Guidance] <br>\n**Output Format:** [Markdown with Python and bash code blocks] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [May request file edits and terminal operations through CellCog Desktop, with write and execute actions subject to user approval.] <br>\n\n## Skill Version(s): <br>\n1.0.13 (source: server release evidence) <br>\n\n## Ethical Considerations: <br>\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment. <br>\n\nArchive v1.0.12: 3 files, 4925 bytes\n\nFiles: skill-card.md (2231b), SKILL.md (8908b), _meta.json (140b)\n\nFile v1.0.12:SKILL.md\n\n---\nname: coding-agent-cellcog\ndescription: \"AI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations — executed directly on your machine. Lightweight with multimedia tools loaded on demand.\"\nauthor: CellCog\nhomepage: https://cellcog.ai\nmetadata:\n  openclaw:\n    emoji: \"💻\"\n    os: [darwin, linux, windows]\n    requires:\n      bins: [python3]\n      env: [CELLCOG_API_KEY]\ndependencies: [cellcog]\n---\n# Coding Agent — The First Coding Agent Built for Agents\n\nWhen your AI needs to code, it delegates to CodeCog. Direct codebase access, terminal operations, and file editing — executed on the user's machine via CellCog Co-work.\n\n## How to Use\n\nFor your first CellCog task in a session, read the **cellcog** skill for the full SDK reference — file handling, chat modes, timeouts, and more.\n\n**OpenClaw (fire-and-forget):**\n```python\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    notify_session_key=\"agent:main:main\",\n    task_label=\"my-task\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\n```\n\n**All agents except OpenClaw (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw|cursor|claude-code|codex|...\")\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    task_label=\"my-task\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\nprint(result[\"message\"])\n```\n\n\n---\n\n## Prerequisites\n\nThis skill requires the `cellcog` skill for SDK setup and API calls.\n\n```bash\nclawhub install cellcog\n```\n\n**Read the cellcog skill first** for SDK setup. This skill shows you how to use CellCog as a coding agent.\n\n**CellCog Desktop Required:** The user must have CellCog Desktop installed and running for Co-work (direct machine access). Download at https://cellcog.ai\n\n---\n\n## Quick Start\n\n**OpenClaw agents (fire-and-forget):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    notify_session_key=\"agent:main:main\",  # OpenClaw only\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**All other agents (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**Key parameters:**\n- `chat_mode=\"agent core\"` — Lightweight coding agent (vs `\"agent\"` for full multimedia)\n- `enable_cowork=True` — Enables Co-work (direct machine access)\n- `cowork_working_directory` — The repo/directory to work in\n\n---\n\n## What CodeCog Can Do\n\n### Code Generation & Editing\n- Write new files, modules, and components\n- Edit existing code with surgical precision\n- Refactor codebases — rename, restructure, extract\n- Port code between languages or frameworks\n\n### Debugging & Fixing\n- Read error logs and stack traces\n- Identify root causes across multiple files\n- Apply fixes and verify they work\n- Run tests to confirm the fix\n\n### Terminal Operations\n- Run build commands, tests, linters\n- Install dependencies (npm, pip, cargo, etc.)\n- Git operations (status, diff, commit)\n- Docker, deployment scripts\n\n### Codebase Exploration\n- Auto-reads AGENTS.md/CLAUDE.md for project conventions\n- Explores directory structure before starting work\n- Understands existing patterns and follows them\n- Reads related files to maintain consistency\n\n---\n\n## What Makes CodeCog Different\n\n### Built for Agents, Not Humans\n\nEvery other coding tool (Cursor, Claude Code, Codex, Windsurf) is designed for human developers sitting at a screen. CodeCog is designed for AI agents that need to code programmatically — fire a request, get results back, continue orchestrating.\n\n### Starts Lean, Scales to Multimodal\n\nCodeCog uses CellCog's Agent Core mode — a lightweight context focused on coding. But if your task unexpectedly needs images, PDFs, videos, or other capabilities, the agent loads those tools on demand. No other coding agent does this.\n\nExample: Your agent asks CodeCog to set up a new project. CodeCog writes the code, then realizes it needs to generate a logo for the README — it loads image tools, generates the logo, and continues. Seamless.\n\n### Direct Machine Access\n\nVia CellCog Co-work, CodeCog operates directly on the user's filesystem:\n- Reads and writes files on the real machine\n- Runs terminal commands in the user's shell\n- Respects project conventions (AGENTS.md, .gitignore, etc.)\n- User approves write/execute operations for safety\n\n---\n\n## Chat Mode\n\n**Always use `\"agent core\"` for CodeCog.** This is the dedicated lightweight mode optimized for coding.\n\n| Mode | Use Case |\n|------|----------|\n| `\"agent core\"` | **CodeCog default** — coding, co-work, terminal ops (50 credits min) |\n| `\"agent\"` | Full multimedia agent — use when you need images/video/audio alongside code (100 credits min) |\n| `\"agent team\"` | Deep research + coding — use for architecture decisions or complex refactors needing research (500 credits min) |\n\n---\n\n## Example Prompts\n\n### New Feature Development\n\n```python\nresult = client.create_chat(\n    prompt=\"Add a REST API endpoint for user profile updates with validation and tests\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"add-profile-api\",\n)\n```\n\n### Bug Fix from Error Log\n\n```python\nresult = client.create_chat(\n    prompt=\"\"\"Fix this error in production:\nTypeError: Cannot read properties of undefined (reading 'map')\nat UserList.render (src/components/UserList.tsx:42)\n\nThe component crashes when the API returns an empty response.\"\"\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"fix-userlist-crash\",\n)\n```\n\n### Codebase Refactor\n\n```python\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module from session-based to JWT tokens. Update all middleware, tests, and API routes.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n### Test Generation\n\n```python\nresult = client.create_chat(\n    prompt=\"Generate comprehensive unit tests for src/services/billing.py. Cover edge cases for proration, currency conversion, and failed payments.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"billing-tests\",\n)\n```\n\nSee https://cellcog.ai for complete SDK API reference — delivery modes, `send_message()`, timeouts, file handling, and more.\n\n---\n\n## Co-work Setup\n\n### Requirements\n1. **CellCog Desktop** must be installed and running on the user's machine\n2. **Working directory** must be specified — this is the root of the project/repo\n3. User must be logged into CellCog Desktop with the same account\n\n### What Co-work Enables\n- `HumanComputer_Terminal` — Run shell commands on the user's machine\n- `HumanComputer_Terminal_File_View` — Read files on the user's machine\n- `HumanComputer_Terminal_File_Write` — Write files on the user's machine\n- `HumanComputer_Terminal_File_Edit` — Edit files on the user's machine\n\n### Safety Model\n- **Read operations** are auto-approved (no interruption)\n- **Write/execute operations** require user approval in the CellCog web UI\n- Users can configure auto-approve for reads/writes within the working directory\n- Sensitive paths (credentials, SSH keys) are always blocked\n\n---\n\n## Tips for Better Results\n\n1. **Specify the working directory** — Always set `cowork_working_directory` to the project root\n2. **Reference specific files** — \"Fix the bug in src/auth/login.ts\" is better than \"fix the login bug\"\n3. **Mention conventions** — \"Follow the existing test patterns\" helps maintain consistency\n4. **Include error context** — Stack traces, log output, and reproduction steps help debugging\n5. **Use AGENTS.md** — Place an AGENTS.md at your repo root with build commands, style guides, and project structure. CodeCog reads it automatically.\n\n---\n\n## Limitations\n\n- **macOS and Linux only** — CellCog Desktop (Co-work) is not yet available on Windows\n- **CellCog Desktop required** — Without Co-work, CodeCog can still write code in its Docker workspace, but cannot access the user's machine directly\n- **User approval for writes** — Write operations pause for user approval (configurable auto-approve available)\n\nFile v1.0.12:_meta.json\n\n{\n  \"ownerId\": \"kn7a96cj9q65e0bhmzahv790en80ffqm\",\n  \"slug\": \"coding-agent-cellcog\",\n  \"version\": \"1.0.12\",\n  \"publishedAt\": 1785703629430\n}\n\nFile v1.0.12:skill-card.md\n\n## Description: <br>\nCoding Agent lets agents delegate code generation, debugging, refactoring, codebase exploration, terminal operations, and file edits to CodeCog through CellCog Co-work on the user's machine. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[cellcog](https://clawhub.ai/user/cellcog) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers and agent builders use this skill when an agent needs to perform coding work in a local project, including generating code, debugging failures, refactoring, running tests, and applying file changes through CellCog Co-work. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: The skill can delegate coding tasks that read files, edit files, and run terminal commands in a project workspace. <br>\nMitigation: Keep write and command approvals manual unless operating in a disposable or well-backed-up directory, and scope the working directory narrowly. <br>\nRisk: Using Co-work gives the coding agent direct local project access when CellCog Desktop is installed and running. <br>\nMitigation: Install this skill only when direct local coding operations are intended, and keep sensitive paths blocked or outside the approved workspace. <br>\n\n\n## Reference(s): <br>\n- [CellCog SDK and Desktop](https://cellcog.ai) <br>\n- [ClawHub Skill Page](https://clawhub.ai/cellcog/skills/coding-agent-cellcog) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [Text, Markdown, Code, Shell commands, Configuration, Guidance] <br>\n**Output Format:** [Markdown responses with code snippets, shell commands, and file-change summaries.] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [May perform local file and terminal actions through CellCog Co-work with user approval for writes and execution.] <br>\n\n## Skill Version(s): <br>\n1.0.12 (source: server release evidence) <br>\n\n## Ethical Considerations: <br>\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment. <br>\n\nArchive v1.0.11: 3 files, 4888 bytes\n\nFiles: skill-card.md (2183b), SKILL.md (8908b), _meta.json (140b)\n\nFile v1.0.11:SKILL.md\n\n---\nname: coding-agent-cellcog\ndescription: \"AI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations — executed directly on your machine. Lightweight with multimedia tools loaded on demand.\"\nauthor: CellCog\nhomepage: https://cellcog.ai\nmetadata:\n  openclaw:\n    emoji: \"💻\"\n    os: [darwin, linux, windows]\n    requires:\n      bins: [python3]\n      env: [CELLCOG_API_KEY]\ndependencies: [cellcog]\n---\n# Coding Agent — The First Coding Agent Built for Agents\n\nWhen your AI needs to code, it delegates to CodeCog. Direct codebase access, terminal operations, and file editing — executed on the user's machine via CellCog Co-work.\n\n## How to Use\n\nFor your first CellCog task in a session, read the **cellcog** skill for the full SDK reference — file handling, chat modes, timeouts, and more.\n\n**OpenClaw (fire-and-forget):**\n```python\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    notify_session_key=\"agent:main:main\",\n    task_label=\"my-task\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\n```\n\n**All agents except OpenClaw (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw|cursor|claude-code|codex|...\")\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    task_label=\"my-task\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\nprint(result[\"message\"])\n```\n\n\n---\n\n## Prerequisites\n\nThis skill requires the `cellcog` skill for SDK setup and API calls.\n\n```bash\nclawhub install cellcog\n```\n\n**Read the cellcog skill first** for SDK setup. This skill shows you how to use CellCog as a coding agent.\n\n**CellCog Desktop Required:** The user must have CellCog Desktop installed and running for Co-work (direct machine access). Download at https://cellcog.ai\n\n---\n\n## Quick Start\n\n**OpenClaw agents (fire-and-forget):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    notify_session_key=\"agent:main:main\",  # OpenClaw only\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**All other agents (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**Key parameters:**\n- `chat_mode=\"agent core\"` — Lightweight coding agent (vs `\"agent\"` for full multimedia)\n- `enable_cowork=True` — Enables Co-work (direct machine access)\n- `cowork_working_directory` — The repo/directory to work in\n\n---\n\n## What CodeCog Can Do\n\n### Code Generation & Editing\n- Write new files, modules, and components\n- Edit existing code with surgical precision\n- Refactor codebases — rename, restructure, extract\n- Port code between languages or frameworks\n\n### Debugging & Fixing\n- Read error logs and stack traces\n- Identify root causes across multiple files\n- Apply fixes and verify they work\n- Run tests to confirm the fix\n\n### Terminal Operations\n- Run build commands, tests, linters\n- Install dependencies (npm, pip, cargo, etc.)\n- Git operations (status, diff, commit)\n- Docker, deployment scripts\n\n### Codebase Exploration\n- Auto-reads AGENTS.md/CLAUDE.md for project conventions\n- Explores directory structure before starting work\n- Understands existing patterns and follows them\n- Reads related files to maintain consistency\n\n---\n\n## What Makes CodeCog Different\n\n### Built for Agents, Not Humans\n\nEvery other coding tool (Cursor, Claude Code, Codex, Windsurf) is designed for human developers sitting at a screen. CodeCog is designed for AI agents that need to code programmatically — fire a request, get results back, continue orchestrating.\n\n### Starts Lean, Scales to Multimodal\n\nCodeCog uses CellCog's Agent Core mode — a lightweight context focused on coding. But if your task unexpectedly needs images, PDFs, videos, or other capabilities, the agent loads those tools on demand. No other coding agent does this.\n\nExample: Your agent asks CodeCog to set up a new project. CodeCog writes the code, then realizes it needs to generate a logo for the README — it loads image tools, generates the logo, and continues. Seamless.\n\n### Direct Machine Access\n\nVia CellCog Co-work, CodeCog operates directly on the user's filesystem:\n- Reads and writes files on the real machine\n- Runs terminal commands in the user's shell\n- Respects project conventions (AGENTS.md, .gitignore, etc.)\n- User approves write/execute operations for safety\n\n---\n\n## Chat Mode\n\n**Always use `\"agent core\"` for CodeCog.** This is the dedicated lightweight mode optimized for coding.\n\n| Mode | Use Case |\n|------|----------|\n| `\"agent core\"` | **CodeCog default** — coding, co-work, terminal ops (50 credits min) |\n| `\"agent\"` | Full multimedia agent — use when you need images/video/audio alongside code (100 credits min) |\n| `\"agent team\"` | Deep research + coding — use for architecture decisions or complex refactors needing research (500 credits min) |\n\n---\n\n## Example Prompts\n\n### New Feature Development\n\n```python\nresult = client.create_chat(\n    prompt=\"Add a REST API endpoint for user profile updates with validation and tests\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"add-profile-api\",\n)\n```\n\n### Bug Fix from Error Log\n\n```python\nresult = client.create_chat(\n    prompt=\"\"\"Fix this error in production:\nTypeError: Cannot read properties of undefined (reading 'map')\nat UserList.render (src/components/UserList.tsx:42)\n\nThe component crashes when the API returns an empty response.\"\"\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"fix-userlist-crash\",\n)\n```\n\n### Codebase Refactor\n\n```python\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module from session-based to JWT tokens. Update all middleware, tests, and API routes.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n### Test Generation\n\n```python\nresult = client.create_chat(\n    prompt=\"Generate comprehensive unit tests for src/services/billing.py. Cover edge cases for proration, currency conversion, and failed payments.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"billing-tests\",\n)\n```\n\nSee https://cellcog.ai for complete SDK API reference — delivery modes, `send_message()`, timeouts, file handling, and more.\n\n---\n\n## Co-work Setup\n\n### Requirements\n1. **CellCog Desktop** must be installed and running on the user's machine\n2. **Working directory** must be specified — this is the root of the project/repo\n3. User must be logged into CellCog Desktop with the same account\n\n### What Co-work Enables\n- `HumanComputer_Terminal` — Run shell commands on the user's machine\n- `HumanComputer_Terminal_File_View` — Read files on the user's machine\n- `HumanComputer_Terminal_File_Write` — Write files on the user's machine\n- `HumanComputer_Terminal_File_Edit` — Edit files on the user's machine\n\n### Safety Model\n- **Read operations** are auto-approved (no interruption)\n- **Write/execute operations** require user approval in the CellCog web UI\n- Users can configure auto-approve for reads/writes within the working directory\n- Sensitive paths (credentials, SSH keys) are always blocked\n\n---\n\n## Tips for Better Results\n\n1. **Specify the working directory** — Always set `cowork_working_directory` to the project root\n2. **Reference specific files** — \"Fix the bug in src/auth/login.ts\" is better than \"fix the login bug\"\n3. **Mention conventions** — \"Follow the existing test patterns\" helps maintain consistency\n4. **Include error context** — Stack traces, log output, and reproduction steps help debugging\n5. **Use AGENTS.md** — Place an AGENTS.md at your repo root with build commands, style guides, and project structure. CodeCog reads it automatically.\n\n---\n\n## Limitations\n\n- **macOS and Linux only** — CellCog Desktop (Co-work) is not yet available on Windows\n- **CellCog Desktop required** — Without Co-work, CodeCog can still write code in its Docker workspace, but cannot access the user's machine directly\n- **User approval for writes** — Write operations pause for user approval (configurable auto-approve available)\n\nFile v1.0.11:_meta.json\n\n{\n  \"ownerId\": \"kn7a96cj9q65e0bhmzahv790en80ffqm\",\n  \"slug\": \"coding-agent-cellcog\",\n  \"version\": \"1.0.11\",\n  \"publishedAt\": 1784735841211\n}\n\nFile v1.0.11:skill-card.md\n\n## Description: <br>\nAI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations - executed directly on your machine. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[nitishgargiitd](https://clawhub.ai/user/nitishgargiitd) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers and agents use this skill to delegate coding tasks to CellCog Co-work for code generation, debugging, refactoring, codebase exploration, file editing, and terminal operations in a selected project directory. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: The skill can let a CellCog-powered agent read, edit, and run commands in a user-selected project directory. <br>\nMitigation: Keep write and execute approvals manual unless the task and repository are trusted, and review proposed changes before relying on them. <br>\nRisk: Using the skill in directories containing secrets or deployment controls may expose or affect sensitive assets. <br>\nMitigation: Choose a scoped working directory and avoid sensitive paths unless the user has carefully reviewed the task and access. <br>\n\n\n## Reference(s): <br>\n- [CellCog](https://cellcog.ai) <br>\n- [ClawHub Skill Page](https://clawhub.ai/nitishgargiitd/skills/coding-agent-cellcog) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [text, markdown, code, shell commands, configuration, guidance] <br>\n**Output Format:** [Markdown or text responses with code blocks, command snippets, and configuration guidance.] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [Requires the CellCog dependency, CELLCOG_API_KEY, and CellCog Desktop for direct Co-work access to a selected project directory.] <br>\n\n## Skill Version(s): <br>\n1.0.11 (source: server release evidence) <br>\n\n## Ethical Considerations: <br>\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment. <br>\n\nArchive v1.0.10: 3 files, 4962 bytes\n\nFiles: skill-card.md (2280b), SKILL.md (8892b), _meta.json (140b)\n\nFile v1.0.10:SKILL.md\n\n---\nname: code-cog\ndescription: \"AI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations — executed directly on your machine. Lightweight with multimedia tools loaded on demand.\"\nauthor: CellCog\nhomepage: https://cellcog.ai\nmetadata:\n  openclaw:\n    emoji: \"💻\"\n    os: [darwin, linux, windows]\n    requires:\n      bins: [python3]\n      env: [CELLCOG_API_KEY]\ndependencies: [cellcog]\n---\n# Code Cog — The First Coding Agent Built for Agents\n\nWhen your AI needs to code, it delegates to CodeCog. Direct codebase access, terminal operations, and file editing — executed on the user's machine via CellCog Co-work.\n\n## How to Use\n\nFor your first CellCog task in a session, read the **cellcog** skill for the full SDK reference — file handling, chat modes, timeouts, and more.\n\n**OpenClaw (fire-and-forget):**\n```python\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    notify_session_key=\"agent:main:main\",\n    task_label=\"my-task\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\n```\n\n**All agents except OpenClaw (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw|cursor|claude-code|codex|...\")\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    task_label=\"my-task\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\nprint(result[\"message\"])\n```\n\n\n---\n\n## Prerequisites\n\nThis skill requires the `cellcog` skill for SDK setup and API calls.\n\n```bash\nclawhub install cellcog\n```\n\n**Read the cellcog skill first** for SDK setup. This skill shows you how to use CellCog as a coding agent.\n\n**CellCog Desktop Required:** The user must have CellCog Desktop installed and running for Co-work (direct machine access). Download at https://cellcog.ai\n\n---\n\n## Quick Start\n\n**OpenClaw agents (fire-and-forget):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    notify_session_key=\"agent:main:main\",  # OpenClaw only\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**All other agents (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**Key parameters:**\n- `chat_mode=\"agent core\"` — Lightweight coding agent (vs `\"agent\"` for full multimedia)\n- `enable_cowork=True` — Enables Co-work (direct machine access)\n- `cowork_working_directory` — The repo/directory to work in\n\n---\n\n## What CodeCog Can Do\n\n### Code Generation & Editing\n- Write new files, modules, and components\n- Edit existing code with surgical precision\n- Refactor codebases — rename, restructure, extract\n- Port code between languages or frameworks\n\n### Debugging & Fixing\n- Read error logs and stack traces\n- Identify root causes across multiple files\n- Apply fixes and verify they work\n- Run tests to confirm the fix\n\n### Terminal Operations\n- Run build commands, tests, linters\n- Install dependencies (npm, pip, cargo, etc.)\n- Git operations (status, diff, commit)\n- Docker, deployment scripts\n\n### Codebase Exploration\n- Auto-reads AGENTS.md/CLAUDE.md for project conventions\n- Explores directory structure before starting work\n- Understands existing patterns and follows them\n- Reads related files to maintain consistency\n\n---\n\n## What Makes CodeCog Different\n\n### Built for Agents, Not Humans\n\nEvery other coding tool (Cursor, Claude Code, Codex, Windsurf) is designed for human developers sitting at a screen. CodeCog is designed for AI agents that need to code programmatically — fire a request, get results back, continue orchestrating.\n\n### Starts Lean, Scales to Multimodal\n\nCodeCog uses CellCog's Agent Core mode — a lightweight context focused on coding. But if your task unexpectedly needs images, PDFs, videos, or other capabilities, the agent loads those tools on demand. No other coding agent does this.\n\nExample: Your agent asks CodeCog to set up a new project. CodeCog writes the code, then realizes it needs to generate a logo for the README — it loads image tools, generates the logo, and continues. Seamless.\n\n### Direct Machine Access\n\nVia CellCog Co-work, CodeCog operates directly on the user's filesystem:\n- Reads and writes files on the real machine\n- Runs terminal commands in the user's shell\n- Respects project conventions (AGENTS.md, .gitignore, etc.)\n- User approves write/execute operations for safety\n\n---\n\n## Chat Mode\n\n**Always use `\"agent core\"` for CodeCog.** This is the dedicated lightweight mode optimized for coding.\n\n| Mode | Use Case |\n|------|----------|\n| `\"agent core\"` | **CodeCog default** — coding, co-work, terminal ops (50 credits min) |\n| `\"agent\"` | Full multimedia agent — use when you need images/video/audio alongside code (100 credits min) |\n| `\"agent team\"` | Deep research + coding — use for architecture decisions or complex refactors needing research (500 credits min) |\n\n---\n\n## Example Prompts\n\n### New Feature Development\n\n```python\nresult = client.create_chat(\n    prompt=\"Add a REST API endpoint for user profile updates with validation and tests\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"add-profile-api\",\n)\n```\n\n### Bug Fix from Error Log\n\n```python\nresult = client.create_chat(\n    prompt=\"\"\"Fix this error in production:\nTypeError: Cannot read properties of undefined (reading 'map')\nat UserList.render (src/components/UserList.tsx:42)\n\nThe component crashes when the API returns an empty response.\"\"\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"fix-userlist-crash\",\n)\n```\n\n### Codebase Refactor\n\n```python\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module from session-based to JWT tokens. Update all middleware, tests, and API routes.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n### Test Generation\n\n```python\nresult = client.create_chat(\n    prompt=\"Generate comprehensive unit tests for src/services/billing.py. Cover edge cases for proration, currency conversion, and failed payments.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"billing-tests\",\n)\n```\n\nSee https://cellcog.ai for complete SDK API reference — delivery modes, `send_message()`, timeouts, file handling, and more.\n\n---\n\n## Co-work Setup\n\n### Requirements\n1. **CellCog Desktop** must be installed and running on the user's machine\n2. **Working directory** must be specified — this is the root of the project/repo\n3. User must be logged into CellCog Desktop with the same account\n\n### What Co-work Enables\n- `HumanComputer_Terminal` — Run shell commands on the user's machine\n- `HumanComputer_Terminal_File_View` — Read files on the user's machine\n- `HumanComputer_Terminal_File_Write` — Write files on the user's machine\n- `HumanComputer_Terminal_File_Edit` — Edit files on the user's machine\n\n### Safety Model\n- **Read operations** are auto-approved (no interruption)\n- **Write/execute operations** require user approval in the CellCog web UI\n- Users can configure auto-approve for reads/writes within the working directory\n- Sensitive paths (credentials, SSH keys) are always blocked\n\n---\n\n## Tips for Better Results\n\n1. **Specify the working directory** — Always set `cowork_working_directory` to the project root\n2. **Reference specific files** — \"Fix the bug in src/auth/login.ts\" is better than \"fix the login bug\"\n3. **Mention conventions** — \"Follow the existing test patterns\" helps maintain consistency\n4. **Include error context** — Stack traces, log output, and reproduction steps help debugging\n5. **Use AGENTS.md** — Place an AGENTS.md at your repo root with build commands, style guides, and project structure. CodeCog reads it automatically.\n\n---\n\n## Limitations\n\n- **macOS and Linux only** — CellCog Desktop (Co-work) is not yet available on Windows\n- **CellCog Desktop required** — Without Co-work, CodeCog can still write code in its Docker workspace, but cannot access the user's machine directly\n- **User approval for writes** — Write operations pause for user approval (configurable auto-approve available)\n\nFile v1.0.10:_meta.json\n\n{\n  \"ownerId\": \"kn7a96cj9q65e0bhmzahv790en80ffqm\",\n  \"slug\": \"coding-agent-cellcog\",\n  \"version\": \"1.0.10\",\n  \"publishedAt\": 1776925116013\n}\n\nFile v1.0.10:skill-card.md\n\n## Description: <br>\nCode Cog lets agents delegate coding tasks to CellCog Co-work for code generation, debugging, refactoring, codebase exploration, and approved local terminal and file operations. <br>\n\nThis skill is ready for commercial/non-commercial use. <br>\n\n## Publisher: <br>\n[nitishgargiitd](https://clawhub.ai/user/nitishgargiitd) <br>\n\n### License/Terms of Use: <br>\nMIT-0 <br>\n\n\n## Use Case: <br>\nDevelopers and agent builders use Code Cog when an agent needs to inspect a project, modify code, run tests or shell commands, and return implementation results through CellCog. <br>\n\n### Deployment Geography for Use: <br>\nGlobal <br>\n\n## Known Risks and Mitigations: <br>\nRisk: Local project file and terminal access can modify files or run commands in the configured workspace. <br>\nMitigation: Keep cowork_working_directory narrow, review terminal and write approvals, and enable write auto-approval only when unattended changes are acceptable. <br>\nRisk: The skill requires a sensitive CELLCOG_API_KEY credential. <br>\nMitigation: Protect the API key and avoid exposing it in prompts, logs, generated files, or shared project artifacts. <br>\nRisk: Generated code or command recommendations may introduce incorrect behavior. <br>\nMitigation: Review diffs and run the project's tests, linters, or build checks before relying on changes. <br>\n\n\n## Reference(s): <br>\n- [Code Cog on ClawHub](https://clawhub.ai/nitishgargiitd/code-cog) <br>\n- [CellCog](https://cellcog.ai) <br>\n\n\n## Skill Output: <br>\n**Output Type(s):** [Text, Markdown, Code, Files, Shell commands, Configuration, Guidance] <br>\n**Output Format:** [Text or Markdown responses with code snippets, shell commands, and generated or modified project files.] <br>\n**Output Parameters:** [1D] <br>\n**Other Properties Related to Output:** [Requires CellCog Desktop, python3, CELLCOG_API_KEY, and a configured working directory for local Co-work access.] <br>\n\n## Skill Version(s): <br>\n1.0.10 (source: server release evidence) <br>\n\n## Ethical Considerations: <br>\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment. <br>\n\nArchive v1.0.9: 2 files, 3699 bytes\n\nFiles: SKILL.md (8818b), _meta.json (139b)\n\nFile v1.0.9:SKILL.md\n\n---\nname: code-cog\ndescription: \"AI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations — executed directly on your machine. Lightweight with multimedia tools loaded on demand.\"\nauthor: CellCog\nhomepage: https://cellcog.ai\nmetadata:\n  openclaw:\n    emoji: \"💻\"\n    os: [darwin, linux]\ndependencies: [cellcog]\n---\n# Code Cog — The First Coding Agent Built for Agents\n\nWhen your AI needs to code, it delegates to CodeCog. Direct codebase access, terminal operations, and file editing — executed on the user's machine via CellCog Co-work.\n\n## How to Use\n\nFor your first CellCog task in a session, read the **cellcog** skill for the full SDK reference — file handling, chat modes, timeouts, and more.\n\n**OpenClaw (fire-and-forget):**\n```python\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    notify_session_key=\"agent:main:main\",\n    task_label=\"my-task\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\n```\n\n**All agents except OpenClaw (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw|cursor|claude-code|codex|...\")\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    task_label=\"my-task\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\nprint(result[\"message\"])\n```\n\n\n---\n\n## Prerequisites\n\nThis skill requires the `cellcog` skill for SDK setup and API calls.\n\n```bash\nclawhub install cellcog\n```\n\n**Read the cellcog skill first** for SDK setup. This skill shows you how to use CellCog as a coding agent.\n\n**CellCog Desktop Required:** The user must have CellCog Desktop installed and running for Co-work (direct machine access). Download at https://cellcog.ai\n\n---\n\n## Quick Start\n\n**OpenClaw agents (fire-and-forget):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    notify_session_key=\"agent:main:main\",  # OpenClaw only\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**All other agents (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**Key parameters:**\n- `chat_mode=\"agent core\"` — Lightweight coding agent (vs `\"agent\"` for full multimedia)\n- `enable_cowork=True` — Enables Co-work (direct machine access)\n- `cowork_working_directory` — The repo/directory to work in\n\n---\n\n## What CodeCog Can Do\n\n### Code Generation & Editing\n- Write new files, modules, and components\n- Edit existing code with surgical precision\n- Refactor codebases — rename, restructure, extract\n- Port code between languages or frameworks\n\n### Debugging & Fixing\n- Read error logs and stack traces\n- Identify root causes across multiple files\n- Apply fixes and verify they work\n- Run tests to confirm the fix\n\n### Terminal Operations\n- Run build commands, tests, linters\n- Install dependencies (npm, pip, cargo, etc.)\n- Git operations (status, diff, commit)\n- Docker, deployment scripts\n\n### Codebase Exploration\n- Auto-reads AGENTS.md/CLAUDE.md for project conventions\n- Explores directory structure before starting work\n- Understands existing patterns and follows them\n- Reads related files to maintain consistency\n\n---\n\n## What Makes CodeCog Different\n\n### Built for Agents, Not Humans\n\nEvery other coding tool (Cursor, Claude Code, Codex, Windsurf) is designed for human developers sitting at a screen. CodeCog is designed for AI agents that need to code programmatically — fire a request, get results back, continue orchestrating.\n\n### Starts Lean, Scales to Multimodal\n\nCodeCog uses CellCog's Agent Core mode — a lightweight context focused on coding. But if your task unexpectedly needs images, PDFs, videos, or other capabilities, the agent loads those tools on demand. No other coding agent does this.\n\nExample: Your agent asks CodeCog to set up a new project. CodeCog writes the code, then realizes it needs to generate a logo for the README — it loads image tools, generates the logo, and continues. Seamless.\n\n### Direct Machine Access\n\nVia CellCog Co-work, CodeCog operates directly on the user's filesystem:\n- Reads and writes files on the real machine\n- Runs terminal commands in the user's shell\n- Respects project conventions (AGENTS.md, .gitignore, etc.)\n- User approves write/execute operations for safety\n\n---\n\n## Chat Mode\n\n**Always use `\"agent core\"` for CodeCog.** This is the dedicated lightweight mode optimized for coding.\n\n| Mode | Use Case |\n|------|----------|\n| `\"agent core\"` | **CodeCog default** — coding, co-work, terminal ops (50 credits min) |\n| `\"agent\"` | Full multimedia agent — use when you need images/video/audio alongside code (100 credits min) |\n| `\"agent team\"` | Deep research + coding — use for architecture decisions or complex refactors needing research (500 credits min) |\n\n---\n\n## Example Prompts\n\n### New Feature Development\n\n```python\nresult = client.create_chat(\n    prompt=\"Add a REST API endpoint for user profile updates with validation and tests\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"add-profile-api\",\n)\n```\n\n### Bug Fix from Error Log\n\n```python\nresult = client.create_chat(\n    prompt=\"\"\"Fix this error in production:\nTypeError: Cannot read properties of undefined (reading 'map')\nat UserList.render (src/components/UserList.tsx:42)\n\nThe component crashes when the API returns an empty response.\"\"\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"fix-userlist-crash\",\n)\n```\n\n### Codebase Refactor\n\n```python\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module from session-based to JWT tokens. Update all middleware, tests, and API routes.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n### Test Generation\n\n```python\nresult = client.create_chat(\n    prompt=\"Generate comprehensive unit tests for src/services/billing.py. Cover edge cases for proration, currency conversion, and failed payments.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"billing-tests\",\n)\n```\n\nSee https://cellcog.ai for complete SDK API reference — delivery modes, `send_message()`, timeouts, file handling, and more.\n\n---\n\n## Co-work Setup\n\n### Requirements\n1. **CellCog Desktop** must be installed and running on the user's machine\n2. **Working directory** must be specified — this is the root of the project/repo\n3. User must be logged into CellCog Desktop with the same account\n\n### What Co-work Enables\n- `HumanComputer_Terminal` — Run shell commands on the user's machine\n- `HumanComputer_Terminal_File_View` — Read files on the user's machine\n- `HumanComputer_Terminal_File_Write` — Write files on the user's machine\n- `HumanComputer_Terminal_File_Edit` — Edit files on the user's machine\n\n### Safety Model\n- **Read operations** are auto-approved (no interruption)\n- **Write/execute operations** require user approval in the CellCog web UI\n- Users can configure auto-approve for reads/writes within the working directory\n- Sensitive paths (credentials, SSH keys) are always blocked\n\n---\n\n## Tips for Better Results\n\n1. **Specify the working directory** — Always set `cowork_working_directory` to the project root\n2. **Reference specific files** — \"Fix the bug in src/auth/login.ts\" is better than \"fix the login bug\"\n3. **Mention conventions** — \"Follow the existing test patterns\" helps maintain consistency\n4. **Include error context** — Stack traces, log output, and reproduction steps help debugging\n5. **Use AGENTS.md** — Place an AGENTS.md at your repo root with build commands, style guides, and project structure. CodeCog reads it automatically.\n\n---\n\n## Limitations\n\n- **macOS and Linux only** — CellCog Desktop (Co-work) is not yet available on Windows\n- **CellCog Desktop required** — Without Co-work, CodeCog can still write code in its Docker workspace, but cannot access the user's machine directly\n- **User approval for writes** — Write operations pause for user approval (configurable auto-approve available)\n\nFile v1.0.9:_meta.json\n\n{\n  \"ownerId\": \"kn7a96cj9q65e0bhmzahv790en80ffqm\",\n  \"slug\": \"coding-agent-cellcog\",\n  \"version\": \"1.0.9\",\n  \"publishedAt\": 1776187909013\n}\n\nArchive v1.0.8: 2 files, 3711 bytes\n\nFiles: SKILL.md (8865b), _meta.json (139b)\n\nFile v1.0.8:SKILL.md\n\n---\nname: code-cog\ndescription: \"The first coding agent built for agents. Code generation, debugging, refactoring, codebase exploration, terminal operations — all executed directly on your machine via CellCog Co-work. Starts lightweight and loads multimedia tools on demand when needed.\"\nauthor: CellCog\nhomepage: https://cellcog.ai\nmetadata:\n  openclaw:\n    emoji: \"💻\"\n    os: [darwin, linux]\ndependencies: [cellcog]\n---\n# Code Cog — The First Coding Agent Built for Agents\n\nWhen your AI needs to code, it delegates to CodeCog. Direct codebase access, terminal operations, and file editing — executed on the user's machine via CellCog Co-work.\n\n## How to Use\n\nFor your first CellCog task in a session, read the **cellcog** skill for the full SDK reference — file handling, chat modes, timeouts, and more.\n\n**OpenClaw (fire-and-forget):**\n```python\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    notify_session_key=\"agent:main:main\",\n    task_label=\"my-task\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\n```\n\n**Cursor / Claude Code / Other agents (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw|cursor|claude-code|codex|...\")\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    task_label=\"my-task\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\nprint(result[\"message\"])\n```\n\n\n---\n\n## Prerequisites\n\nThis skill requires the `cellcog` skill for SDK setup and API calls.\n\n```bash\nclawhub install cellcog\n```\n\n**Read the cellcog skill first** for SDK setup. This skill shows you how to use CellCog as a coding agent.\n\n**CellCog Desktop Required:** The user must have CellCog Desktop installed and running for Co-work (direct machine access). Download at https://cellcog.ai\n\n---\n\n## Quick Start\n\n**OpenClaw agents (fire-and-forget):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    notify_session_key=\"agent:main:main\",  # OpenClaw only\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**All other agents (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**Key parameters:**\n- `chat_mode=\"agent core\"` — Lightweight coding agent (vs `\"agent\"` for full multimedia)\n- `enable_cowork=True` — Enables Co-work (direct machine access)\n- `cowork_working_directory` — The repo/directory to work in\n\n---\n\n## What CodeCog Can Do\n\n### Code Generation & Editing\n- Write new files, modules, and components\n- Edit existing code with surgical precision\n- Refactor codebases — rename, restructure, extract\n- Port code between languages or frameworks\n\n### Debugging & Fixing\n- Read error logs and stack traces\n- Identify root causes across multiple files\n- Apply fixes and verify they work\n- Run tests to confirm the fix\n\n### Terminal Operations\n- Run build commands, tests, linters\n- Install dependencies (npm, pip, cargo, etc.)\n- Git operations (status, diff, commit)\n- Docker, deployment scripts\n\n### Codebase Exploration\n- Auto-reads AGENTS.md/CLAUDE.md for project conventions\n- Explores directory structure before starting work\n- Understands existing patterns and follows them\n- Reads related files to maintain consistency\n\n---\n\n## What Makes CodeCog Different\n\n### Built for Agents, Not Humans\n\nEvery other coding tool (Cursor, Claude Code, Codex, Windsurf) is designed for human developers sitting at a screen. CodeCog is designed for AI agents that need to code programmatically — fire a request, get results back, continue orchestrating.\n\n### Starts Lean, Scales to Multimodal\n\nCodeCog uses CellCog's Agent Core mode — a lightweight context focused on coding. But if your task unexpectedly needs images, PDFs, videos, or other capabilities, the agent loads those tools on demand. No other coding agent does this.\n\nExample: Your agent asks CodeCog to set up a new project. CodeCog writes the code, then realizes it needs to generate a logo for the README — it loads image tools, generates the logo, and continues. Seamless.\n\n### Direct Machine Access\n\nVia CellCog Co-work, CodeCog operates directly on the user's filesystem:\n- Reads and writes files on the real machine\n- Runs terminal commands in the user's shell\n- Respects project conventions (AGENTS.md, .gitignore, etc.)\n- User approves write/execute operations for safety\n\n---\n\n## Chat Mode\n\n**Always use `\"agent core\"` for CodeCog.** This is the dedicated lightweight mode optimized for coding.\n\n| Mode | Use Case |\n|------|----------|\n| `\"agent core\"` | **CodeCog default** — coding, co-work, terminal ops (50 credits min) |\n| `\"agent\"` | Full multimedia agent — use when you need images/video/audio alongside code (100 credits min) |\n| `\"agent team\"` | Deep research + coding — use for architecture decisions or complex refactors needing research (500 credits min) |\n\n---\n\n## Example Prompts\n\n### New Feature Development\n\n```python\nresult = client.create_chat(\n    prompt=\"Add a REST API endpoint for user profile updates with validation and tests\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"add-profile-api\",\n)\n```\n\n### Bug Fix from Error Log\n\n```python\nresult = client.create_chat(\n    prompt=\"\"\"Fix this error in production:\nTypeError: Cannot read properties of undefined (reading 'map')\nat UserList.render (src/components/UserList.tsx:42)\n\nThe component crashes when the API returns an empty response.\"\"\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"fix-userlist-crash\",\n)\n```\n\n### Codebase Refactor\n\n```python\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module from session-based to JWT tokens. Update all middleware, tests, and API routes.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n### Test Generation\n\n```python\nresult = client.create_chat(\n    prompt=\"Generate comprehensive unit tests for src/services/billing.py. Cover edge cases for proration, currency conversion, and failed payments.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"billing-tests\",\n)\n```\n\nSee https://cellcog.ai for complete SDK API reference — delivery modes, `send_message()`, timeouts, file handling, and more.\n\n---\n\n## Co-work Setup\n\n### Requirements\n1. **CellCog Desktop** must be installed and running on the user's machine\n2. **Working directory** must be specified — this is the root of the project/repo\n3. User must be logged into CellCog Desktop with the same account\n\n### What Co-work Enables\n- `HumanComputer_Terminal` — Run shell commands on the user's machine\n- `HumanComputer_Terminal_File_View` — Read files on the user's machine\n- `HumanComputer_Terminal_File_Write` — Write files on the user's machine\n- `HumanComputer_Terminal_File_Edit` — Edit files on the user's machine\n\n### Safety Model\n- **Read operations** are auto-approved (no interruption)\n- **Write/execute operations** require user approval in the CellCog web UI\n- Users can configure auto-approve for reads/writes within the working directory\n- Sensitive paths (credentials, SSH keys) are always blocked\n\n---\n\n## Tips for Better Results\n\n1. **Specify the working directory** — Always set `cowork_working_directory` to the project root\n2. **Reference specific files** — \"Fix the bug in src/auth/login.ts\" is better than \"fix the login bug\"\n3. **Mention conventions** — \"Follow the existing test patterns\" helps maintain consistency\n4. **Include error context** — Stack traces, log output, and reproduction steps help debugging\n5. **Use AGENTS.md** — Place an AGENTS.md at your repo root with build commands, style guides, and project structure. CodeCog reads it automatically.\n\n---\n\n## Limitations\n\n- **macOS and Linux only** — CellCog Desktop (Co-work) is not yet available on Windows\n- **CellCog Desktop required** — Without Co-work, CodeCog can still write code in its Docker workspace, but cannot access the user's machine directly\n- **User approval for writes** — Write operations pause for user approval (configurable auto-approve available)\n\nFile v1.0.8:_meta.json\n\n{\n  \"ownerId\": \"kn7a96cj9q65e0bhmzahv790en80ffqm\",\n  \"slug\": \"coding-agent-cellcog\",\n  \"version\": \"1.0.8\",\n  \"publishedAt\": 1776042025534\n}\n\nArchive v1.0.7: 2 files, 3656 bytes\n\nFiles: SKILL.md (8693b), _meta.json (139b)\n\nFile v1.0.7:SKILL.md\n\n---\nname: code-cog\ndescription: \"Powered by CellCog. Coding agent for agents. Code generation, debugging, refactoring, codebase exploration, and terminal operations — executed directly on the user's machine via CellCog Co-work.\"\nauthor: CellCog\nhomepage: https://cellcog.ai\nmetadata:\n  openclaw:\n    emoji: \"💻\"\n    os: [darwin, linux]\ndependencies: [cellcog]\n---\n# Code Cog — The First Coding Agent Built for Agents\n\nWhen your AI needs to code, it delegates to CodeCog. Direct codebase access, terminal operations, and file editing — executed on the user's machine via CellCog Co-work.\n\n## How to Use\n\nFor your first CellCog task in a session, read the **cellcog** skill for the full SDK reference — file handling, chat modes, timeouts, and more.\n\n**OpenClaw (fire-and-forget):**\n```python\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    notify_session_key=\"agent:main:main\",\n    task_label=\"my-task\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\n```\n\n**Cursor / Claude Code / Other agents (blocks until done):**\n```python\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    task_label=\"my-task\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\nprint(result[\"message\"])\n```\n\n\n---\n\n## Prerequisites\n\nThis skill requires the `cellcog` skill for SDK setup and API calls.\n\n```bash\nclawhub install cellcog\n```\n\n**Read the cellcog skill first** for SDK setup. This skill shows you how to use CellCog as a coding agent.\n\n**CellCog Desktop Required:** The user must have CellCog Desktop installed and running for Co-work (direct machine access). Download at https://cellcog.ai\n\n---\n\n## Quick Start\n\n**OpenClaw agents (fire-and-forget):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    notify_session_key=\"agent:main:main\",  # OpenClaw only\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**All other agents (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**Key parameters:**\n- `chat_mode=\"agent core\"` — Lightweight coding agent (vs `\"agent\"` for full multimedia)\n- `enable_cowork=True` — Enables Co-work (direct machine access)\n- `cowork_working_directory` — The repo/directory to work in\n\n---\n\n## What CodeCog Can Do\n\n### Code Generation & Editing\n- Write new files, modules, and components\n- Edit existing code with surgical precision\n- Refactor codebases — rename, restructure, extract\n- Port code between languages or frameworks\n\n### Debugging & Fixing\n- Read error logs and stack traces\n- Identify root causes across multiple files\n- Apply fixes and verify they work\n- Run tests to confirm the fix\n\n### Terminal Operations\n- Run build commands, tests, linters\n- Install dependencies (npm, pip, cargo, etc.)\n- Git operations (status, diff, commit)\n- Docker, deployment scripts\n\n### Codebase Exploration\n- Auto-reads AGENTS.md/CLAUDE.md for project conventions\n- Explores directory structure before starting work\n- Understands existing patterns and follows them\n- Reads related files to maintain consistency\n\n---\n\n## What Makes CodeCog Different\n\n### Built for Agents, Not Humans\n\nEvery other coding tool (Cursor, Claude Code, Codex, Windsurf) is designed for human developers sitting at a screen. CodeCog is designed for AI agents that need to code programmatically — fire a request, get results back, continue orchestrating.\n\n### Starts Lean, Scales to Multimodal\n\nCodeCog uses CellCog's Agent Core mode — a lightweight context focused on coding. But if your task unexpectedly needs images, PDFs, videos, or other capabilities, the agent loads those tools on demand. No other coding agent does this.\n\nExample: Your agent asks CodeCog to set up a new project. CodeCog writes the code, then realizes it needs to generate a logo for the README — it loads image tools, generates the logo, and continues. Seamless.\n\n### Direct Machine Access\n\nVia CellCog Co-work, CodeCog operates directly on the user's filesystem:\n- Reads and writes files on the real machine\n- Runs terminal commands in the user's shell\n- Respects project conventions (AGENTS.md, .gitignore, etc.)\n- User approves write/execute operations for safety\n\n---\n\n## Chat Mode\n\n**Always use `\"agent core\"` for CodeCog.** This is the dedicated lightweight mode optimized for coding.\n\n| Mode | Use Case |\n|------|----------|\n| `\"agent core\"` | **CodeCog default** — coding, co-work, terminal ops (50 credits min) |\n| `\"agent\"` | Full multimedia agent — use when you need images/video/audio alongside code (100 credits min) |\n| `\"agent team\"` | Deep research + coding — use for architecture decisions or complex refactors needing research (500 credits min) |\n\n---\n\n## Example Prompts\n\n### New Feature Development\n\n```python\nresult = client.create_chat(\n    prompt=\"Add a REST API endpoint for user profile updates with validation and tests\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"add-profile-api\",\n)\n```\n\n### Bug Fix from Error Log\n\n```python\nresult = client.create_chat(\n    prompt=\"\"\"Fix this error in production:\nTypeError: Cannot read properties of undefined (reading 'map')\nat UserList.render (src/components/UserList.tsx:42)\n\nThe component crashes when the API returns an empty response.\"\"\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"fix-userlist-crash\",\n)\n```\n\n### Codebase Refactor\n\n```python\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module from session-based to JWT tokens. Update all middleware, tests, and API routes.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n### Test Generation\n\n```python\nresult = client.create_chat(\n    prompt=\"Generate comprehensive unit tests for src/services/billing.py. Cover edge cases for proration, currency conversion, and failed payments.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"billing-tests\",\n)\n```\n\nSee https://cellcog.ai for complete SDK API reference — delivery modes, `send_message()`, timeouts, file handling, and more.\n\n---\n\n## Co-work Setup\n\n### Requirements\n1. **CellCog Desktop** must be installed and running on the user's machine\n2. **Working directory** must be specified — this is the root of the project/repo\n3. User must be logged into CellCog Desktop with the same account\n\n### What Co-work Enables\n- `HumanComputer_Terminal` — Run shell commands on the user's machine\n- `HumanComputer_Terminal_File_View` — Read files on the user's machine\n- `HumanComputer_Terminal_File_Write` — Write files on the user's machine\n- `HumanComputer_Terminal_File_Edit` — Edit files on the user's machine\n\n### Safety Model\n- **Read operations** are auto-approved (no interruption)\n- **Write/execute operations** require user approval in the CellCog web UI\n- Users can configure auto-approve for reads/writes within the working directory\n- Sensitive paths (credentials, SSH keys) are always blocked\n\n---\n\n## Tips for Better Results\n\n1. **Specify the working directory** — Always set `cowork_working_directory` to the project root\n2. **Reference specific files** — \"Fix the bug in src/auth/login.ts\" is better than \"fix the login bug\"\n3. **Mention conventions** — \"Follow the existing test patterns\" helps maintain consistency\n4. **Include error context** — Stack traces, log output, and reproduction steps help debugging\n5. **Use AGENTS.md** — Place an AGENTS.md at your repo root with build commands, style guides, and project structure. CodeCog reads it automatically.\n\n---\n\n## Limitations\n\n- **macOS and Linux only** — CellCog Desktop (Co-work) is not yet available on Windows\n- **CellCog Desktop required** — Without Co-work, CodeCog can still write code in its Docker workspace, but cannot access the user's machine directly\n- **User approval for writes** — Write operations pause for user approval (configurable auto-approve available)\n\nFile v1.0.7:_meta.json\n\n{\n  \"ownerId\": \"kn7a96cj9q65e0bhmzahv790en80ffqm\",\n  \"slug\": \"coding-agent-cellcog\",\n  \"version\": \"1.0.7\",\n  \"publishedAt\": 1776036901360\n}\n\nArchive v1.0.6: 2 files, 3557 bytes\n\nFiles: SKILL.md (8031b), _meta.json (139b)\n\nFile v1.0.6:SKILL.md\n\n---\nname: code-cog\ndescription: \"The first coding agent built for agents. Code generation, debugging, refactoring, codebase exploration, terminal operations — all executed directly on your machine via CellCog Co-work. Starts lightweight and loads multimedia tools on demand when needed.\"\nauthor: CellCog\nhomepage: https://cellcog.ai\nmetadata:\n  openclaw:\n    emoji: \"💻\"\n    os: [darwin, linux]\ndependencies: [cellcog]\n---\n\n# Code Cog — The First Coding Agent Built for Agents\n\nWhen your AI needs to code, it delegates to CodeCog. Direct codebase access, terminal operations, and file editing — executed on the user's machine via CellCog Co-work.\n\n---\n\n## Prerequisites\n\nThis skill requires the `cellcog` skill for SDK setup and API calls.\n\n```bash\nclawhub install cellcog\n```\n\n**Read the cellcog skill first** for SDK setup. This skill shows you how to use CellCog as a coding agent.\n\n**CellCog Desktop Required:** The user must have CellCog Desktop installed and running for Co-work (direct machine access). Download at https://cellcog.ai\n\n---\n\n## Quick Start\n\n**OpenClaw agents (fire-and-forget):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    notify_session_key=\"agent:main:main\",  # OpenClaw only\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**All other agents (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**Key parameters:**\n- `chat_mode=\"agent core\"` — Lightweight coding agent (vs `\"agent\"` for full multimedia)\n- `enable_cowork=True` — Enables Co-work (direct machine access)\n- `cowork_working_directory` — The repo/directory to work in\n\n---\n\n## What CodeCog Can Do\n\n### Code Generation & Editing\n- Write new files, modules, and components\n- Edit existing code with surgical precision\n- Refactor codebases — rename, restructure, extract\n- Port code between languages or frameworks\n\n### Debugging & Fixing\n- Read error logs and stack traces\n- Identify root causes across multiple files\n- Apply fixes and verify they work\n- Run tests to confirm the fix\n\n### Terminal Operations\n- Run build commands, tests, linters\n- Install dependencies (npm, pip, cargo, etc.)\n- Git operations (status, diff, commit)\n- Docker, deployment scripts\n\n### Codebase Exploration\n- Auto-reads AGENTS.md/CLAUDE.md for project conventions\n- Explores directory structure before starting work\n- Understands existing patterns and follows them\n- Reads related files to maintain consistency\n\n---\n\n## What Makes CodeCog Different\n\n### Built for Agents, Not Humans\n\nEvery other coding tool (Cursor, Claude Code, Codex, Windsurf) is designed for human developers sitting at a screen. CodeCog is designed for AI agents that need to code programmatically — fire a request, get results back, continue orchestrating.\n\n### Starts Lean, Scales to Multimodal\n\nCodeCog uses CellCog's Agent Core mode — a lightweight context focused on coding. But if your task unexpectedly needs images, PDFs, videos, or other capabilities, the agent loads those tools on demand. No other coding agent does this.\n\nExample: Your agent asks CodeCog to set up a new project. CodeCog writes the code, then realizes it needs to generate a logo for the README — it loads image tools, generates the logo, and continues. Seamless.\n\n### Direct Machine Access\n\nVia CellCog Co-work, CodeCog operates directly on the user's filesystem:\n- Reads and writes files on the real machine\n- Runs terminal commands in the user's shell\n- Respects project conventions (AGENTS.md, .gitignore, etc.)\n- User approves write/execute operations for safety\n\n---\n\n## Chat Mode\n\n**Always use `\"agent core\"` for CodeCog.** This is the dedicated lightweight mode optimized for coding.\n\n| Mode | Use Case |\n|------|----------|\n| `\"agent core\"` | **CodeCog default** — coding, co-work, terminal ops (50 credits min) |\n| `\"agent\"` | Full multimedia agent — use when you need images/video/audio alongside code (100 credits min) |\n| `\"agent team\"` | Deep research + coding — use for architecture decisions or complex refactors needing research (500 credits min) |\n\n---\n\n## Example Prompts\n\n### New Feature Development\n\n```python\nresult = client.create_chat(\n    prompt=\"Add a REST API endpoint for user profile updates with validation and tests\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"add-profile-api\",\n)\n```\n\n### Bug Fix from Error Log\n\n```python\nresult = client.create_chat(\n    prompt=\"\"\"Fix this error in production:\nTypeError: Cannot read properties of undefined (reading 'map')\nat UserList.render (src/components/UserList.tsx:42)\n\nThe component crashes when the API returns an empty response.\"\"\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"fix-userlist-crash\",\n)\n```\n\n### Codebase Refactor\n\n```python\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module from session-based to JWT tokens. Update all middleware, tests, and API routes.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n### Test Generation\n\n```python\nresult = client.create_chat(\n    prompt=\"Generate comprehensive unit tests for src/services/billing.py. Cover edge cases for proration, currency conversion, and failed payments.\",\n    chat_mode=\"agent core\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"billing-tests\",\n)\n```\n\nSee the **cellcog** mothership skill for complete SDK API reference — delivery modes, `send_message()`, timeouts, file handling, and more.\n\n---\n\n## Co-work Setup\n\n### Requirements\n1. **CellCog Desktop** must be installed and running on the user's machine\n2. **Working directory** must be specified — this is the root of the project/repo\n3. User must be logged into CellCog Desktop with the same account\n\n### What Co-work Enables\n- `HumanComputer_Terminal` — Run shell commands on the user's machine\n- `HumanComputer_Terminal_File_View` — Read files on the user's machine\n- `HumanComputer_Terminal_File_Write` — Write files on the user's machine\n- `HumanComputer_Terminal_File_Edit` — Edit files on the user's machine\n\n### Safety Model\n- **Read operations** are auto-approved (no interruption)\n- **Write/execute operations** require user approval in the CellCog web UI\n- Users can configure auto-approve for reads/writes within the working directory\n- Sensitive paths (credentials, SSH keys) are always blocked\n\n---\n\n## Tips for Better Results\n\n1. **Specify the working directory** — Always set `cowork_working_directory` to the project root\n2. **Reference specific files** — \"Fix the bug in src/auth/login.ts\" is better than \"fix the login bug\"\n3. **Mention conventions** — \"Follow the existing test patterns\" helps maintain consistency\n4. **Include error context** — Stack traces, log output, and reproduction steps help debugging\n5. **Use AGENTS.md** — Place an AGENTS.md at your repo root with build commands, style guides, and project structure. CodeCog reads it automatically.\n\n---\n\n## Limitations\n\n- **macOS and Linux only** — CellCog Desktop (Co-work) is not yet available on Windows\n- **CellCog Desktop required** — Without Co-work, CodeCog can still write code in its Docker workspace, but cannot access the user's machine directly\n- **User approval for writes** — Write operations pause for user approval (configurable auto-approve available)\n\nFile v1.0.6:_meta.json\n\n{\n  \"ownerId\": \"kn7a96cj9q65e0bhmzahv790en80ffqm\",\n  \"slug\": \"coding-agent-cellcog\",\n  \"version\": \"1.0.6\",\n  \"publishedAt\": 1775886748003\n}","readmeExcerpt":"Skill: Coding Agent Owner: cellcog Summary: AI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations — executed directly on your machine. Lightweight with multimedia tools loaded on demand. Tags: latest:1.0.15 Version history: v1.0.15 | 2026-08-24T02:01:28.499Z | user Content updated. v1.0.14 | 2026-08-24T01:45:50.351Z | user Content updated. v1.0.","codeSnippets":[],"executableExamples":[{"language":"python","snippet":"result = client.create_chat(\n    prompt=\"[your task prompt]\",\n    notify_session_key=\"agent:main:main\",\n    task_label=\"my-task\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)"},{"language":"python","snippet":"from cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw|cursor|claude-code|codex|...\")\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    task_label=\"my-task\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\nprint(result[\"message\"])"},{"language":"bash","snippet":"# Claude Code, Cursor, Codex + 70 more agents\nnpx skills add cellcog/skills --skill cellcog\n\n# OpenClaw\nopenclaw skills install @cellcog/cellcog"},{"language":"python","snippet":"from cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    notify_session_key=\"agent:main:main\",  # OpenClaw only\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)"},{"language":"python","snippet":"from cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)"},{"language":"python","snippet":"result = client.create_chat(\n    prompt=\"Add a REST API endpoint for user profile updates with validation and tests\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"add-profile-api\",\n)"}],"parameters":null,"dependencies":[],"permissions":[],"extractedFiles":[{"path":"SKILL.md","content":"---\nname: coding-agent-cellcog\ndescription: \"AI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations — executed directly on your machine. Lightweight with multimedia tools loaded on demand.\"\nauthor: CellCog\nhomepage: https://cellcog.ai\nmetadata:\n  openclaw:\n    emoji: \"💻\"\n    os: [darwin, linux, windows]\n    requires:\n      bins: [python3]\n      env: [CELLCOG_API_KEY]\ndependencies: [cellcog]\n---\n# Coding Agent — The First Coding Agent Built for Agents\n\nWhen your AI needs to code, it delegates to CodeCog. Direct codebase access, terminal operations, and file editing — executed on the user's machine via CellCog Co-work.\n\n## How to Use\n\nFor your first CellCog task in a session, read the **cellcog** skill for the full SDK reference — file handling, chat modes, timeouts, and more.\n\n**OpenClaw (fire-and-forget):**\n```python\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    notify_session_key=\"agent:main:main\",\n    task_label=\"my-task\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\n```\n\n**All agents except OpenClaw (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw|cursor|claude-code|codex|...\")\nresult = client.create_chat(\n    prompt=\"[your task prompt]\",\n    task_label=\"my-task\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/path/to/project\",\n)\nprint(result[\"message\"])\n```\n\n\n---\n\n## Prerequisites\n\nThis skill requires the `cellcog` skill for SDK setup and API calls.\n\n```bash\n# Claude Code, Cursor, Codex + 70 more agents\nnpx skills add cellcog/skills --skill cellcog\n\n# OpenClaw\nopenclaw skills install @cellcog/cellcog\n```\n\n**Read the cellcog skill first** for SDK setup. This skill shows you how to use CellCog as a coding agent.\n\n**CellCog Desktop Required:** The user must have CellCog Desktop installed and running for Co-work (direct machine access). Download at https://cellcog.ai\n\n---\n\n## Quick Start\n\n**OpenClaw agents (fire-and-forget):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    notify_session_key=\"agent:main:main\",  # OpenClaw only\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**All other agents (blocks until done):**\n```python\nfrom cellcog import CellCogClient\nclient = CellCogClient(agent_provider=\"openclaw\")\n\nresult = client.create_chat(\n    prompt=\"Refactor the authentication module to use JWT tokens\",\n    chat_mode=\"agent\",\n    chat_tier=\"max\",\n    enable_cowork=True,\n    cowork_working_directory=\"/Users/me/projects/myapp\",\n    task_label=\"auth-refactor\",\n)\n```\n\n**Key parameters:**\n- `chat_mode=\"agent\", cha"},{"path":"_meta.json","content":"{\n  \"ownerId\": \"kn7a96cj9q65e0bhmzahv790en80ffqm\",\n  \"slug\": \"coding-agent-cellcog\",\n  \"version\": \"1.0.15\",\n  \"publishedAt\": 1787536888499\n}"},{"path":"skill-card.md","content":"## Description:\n\nAI coding agent powered by CellCog Co-work for code generation, debugging, refactoring, codebase exploration, and terminal operations executed on the user's machine.\n\nThis skill is ready for commercial/non-commercial use.\n\n## Publisher:\n\n[cellcog](https://clawhub.ai/user/cellcog)\n\n### License/Terms of Use:\n\nMIT-0\n\n## Use Case:\n\nDevelopers and agents use this skill to delegate coding work to CellCog Co-work, including code generation, bug fixing, refactoring, repository exploration, tests, and terminal operations in a specified project directory.\n\n### Deployment Geography for Use:\n\nGlobal\n\n## Known Risks and Mitigations:\n\nRisk: CellCog Co-work gives a coding agent local project access for file reads, file writes, and terminal commands.\n\nMitigation: Scope Co-work to a single project directory and review write and command actions before approval.\n\nRisk: Configurable write auto-approval can allow broader changes inside the selected working directory.\n\nMitigation: Leave write auto-approve disabled unless the installed CellCog components and project scope have been verified.\n\nRisk: The skill depends on separately installed CellCog Desktop/runtime components.\n\nMitigation: Install only if you trust CellCog and have verified the installed CellCog versions.\n\n## Reference(s):\n\n- [CellCog](https://cellcog.ai)\n- [ClawHub Coding Agent Skill](https://clawhub.ai/cellcog/skills/coding-agent-cellcog)\n\n## Skill Output:\n\n**Output Type(s):** [Text, Markdown, Code, Shell commands, Configuration, Guidance]\n\n**Output Format:** [Markdown and text with inline code, shell commands, and configuration snippets]\n\n**Output Parameters:** [1D]\n\n**Other Properties Related to Output:** [May produce file edits and terminal actions through CellCog Co-work subject to the user's local approval settings.]\n\n## Skill Version(s):\n\n1.0.15 (source: server release metadata)\n\n## Ethical Considerations:\n\nUsers should evaluate whether this skill is appropriate for their environment, review any generated or modified files before relying on them, and apply their organization's safety, security, and compliance requirements before deployment."}],"languages":[],"docsSourceLabel":"CLAWHUB","editorialOverview":"AI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations — executed directly on your machine. Lightweight with multimedia tools loaded on demand. Skill: Coding Agent Owner: cellcog Summary: AI coding agent powered by CellCog Co-work. Code generation, debugging, refactoring, codebase exploration, terminal operations — executed directly on your machine. Lightweight with multimedia tools loaded on demand. Tags: latest:1.0.15 Version history: v1.0.15 | 2026-08-24T02:01:28.499Z | user Content updated. v1.0.14 | 2026-08-24T01:45:50.351Z | user Content updated. v1.0.","editorialQuality":{"score":100,"threshold":65,"status":"ready","wordCount":1076,"uniquenessScore":50,"reasons":[]}},"media":{"evidence":{"source":"no-media","verified":false,"confidence":"low","updatedAt":"2026-10-09T15:52:35.177Z","emptyReason":"No screenshots, media assets, or demo links are available."},"primaryImageUrl":null,"mediaAssetCount":0,"assets":[],"demoUrl":null},"ownerResources":{"evidence":{"source":"unclaimed","verified":false,"confidence":"low","updatedAt":"2026-10-09T15:52:35.177Z","emptyReason":"This page has not been claimed by the agent owner."},"hasCustomPage":false,"customPageUpdatedAt":null,"customLinks":[],"structuredLinks":{"docsUrl":null,"demoUrl":null,"supportUrl":null,"pricingUrl":null,"statusUrl":null},"customPage":null},"relatedAgents":{"evidence":{"source":"protocol-neighbors","verified":false,"confidence":"medium","updatedAt":"2026-10-10T03:51:29.070Z","emptyReason":null},"items":[{"id":"8ebccd8e-3863-4187-8355-c3f14e1f9edf","entityType":"agent","canonicalPath":"/agent/iofficeai-aionui","slug":"iofficeai-aionui","name":"AionUi","description":"Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!","url":"https://github.com/iOfficeAI/AionUi","homepage":"https://www.aionui.com","source":"GITHUB_REPOS","protocols":["MCP","OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-10-09T19:11:12.944Z","createdAt":"2026-02-25T03:38:16.584Z","downloads":null},{"id":"b917f68a-ebff-438e-84f8-3f4b2494c0bc","entityType":"agent","canonicalPath":"/agent/activepieces-activepieces","slug":"activepieces-activepieces","name":"activepieces","description":"AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents","url":"https://github.com/activepieces/activepieces","homepage":"https://www.activepieces.com","source":"GITHUB_REPOS","protocols":["OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-04-15T02:22:12.426Z","createdAt":"2026-02-25T03:38:12.412Z","downloads":null},{"id":"5cb26759-3a39-483f-94cf-276a98c13bb8","entityType":"agent","canonicalPath":"/agent/cherryhq-cherry-studio","slug":"cherryhq-cherry-studio","name":"cherry-studio","description":"AI productivity studio with smart chat, autonomous agents, and 300+ assistants. Unified access to frontier LLMs","url":"https://github.com/CherryHQ/cherry-studio","homepage":"https://cherry-ai.com","source":"GITHUB_REPOS","protocols":["MCP","OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-04-11T14:38:40.986Z","createdAt":"2026-02-25T03:38:19.379Z","downloads":null},{"id":"6f6582d0-5d76-4f0f-b81d-86520247950b","entityType":"agent","canonicalPath":"/agent/copilotkit-copilotkit","slug":"copilotkit-copilotkit","name":"CopilotKit","description":"The Frontend for Agents & Generative UI. React + Angular","url":"https://github.com/CopilotKit/CopilotKit","homepage":"https://docs.copilotkit.ai","source":"GITHUB_REPOS","protocols":["OPENCLAW"],"capabilities":[],"safetyScore":100,"overallRank":70,"updatedAt":"2026-03-25T09:50:57.846Z","createdAt":"2026-02-25T03:39:14.617Z","downloads":null}],"links":{"hub":"/agent","source":"/agent/source/clawhub","protocols":[{"label":"OpenClaw","href":"/agent/protocol/openclew"}]}}}