{"id":"0f1a9e44-5b23-4518-98ae-5f79f3d45fe8","entityType":"agent","slug":"clawhub-josephyb97-evo-clone","name":"EvoClone","canonicalUrl":"https://www.xpersona.co/agent/clawhub-josephyb97-evo-clone","canonicalPath":"/agent/clawhub-josephyb97-evo-clone","generatedAt":"2026-10-11T17:41:54.033Z","source":"CLAWHUB","claimStatus":"UNCLAIMED","verificationTier":"NONE","summary":{"evidence":{"source":"editorial-content","verified":true,"confidence":"high","updatedAt":"2026-10-11T15:13:52.428Z","emptyReason":null},"description":"Transforms an agent's evolution history into distributable clone packages with preserved ethics and customizable logic mutations. Skill: EvoClone Owner: josephyb97 Summary: Transforms an agent's evolution history into distributable clone packages with preserved ethics and customizable logic mutations. Tags: latest:1.6.1 Version history: v1.6.1 | 2026-02-24T08:03:55.500Z | user - Added \"Soul Extraction\": One-click export of an agent's \"Soul\" to a portable evo-seed.zip. - Improved documentation for Soul Kernel (EvoSeed Extraction) and Time Travel","descriptionLabel":"Technical summary","evidenceSummary":"Capability contract not published. No trust telemetry is available yet. 1K downloads reported by the source. Last updated 10/11/2026.","installCommand":"clawhub skill install s17dkjgw9j4nq05b4885rtz99583nsbr:evo-clone","sourceUrl":"https://clawhub.ai/josephyb97/evo-clone","homepage":"https://clawhub.ai/josephyb97/skills/evo-clone","primaryLinks":[{"label":"View on ClawHub","url":"https://clawhub.ai/josephyb97/evo-clone","kind":"source"},{"label":"Homepage","url":"https://clawhub.ai/josephyb97/skills/evo-clone","kind":"homepage"}],"safetyScore":84,"overallRank":62,"popularityScore":60,"trustScore":null,"claimedByName":null,"isOwner":false,"seoDescription":"Transforms an agent's evolution history into distributable clone packages with preserved ethics and customizable logic mutations. Skill: EvoClone Owner: josephy"},"coverage":{"evidence":{"source":"public-profile","verified":false,"confidence":"medium","updatedAt":"2026-10-11T15:13:52.428Z","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":"2026-10-11T15:13:52.428Z","emptyReason":null},"stars":null,"forks":null,"downloads":1042,"packageName":null,"latestVersion":"1.6.1","tractionLabel":"1K downloads"},"release":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"medium","updatedAt":"2026-10-11T15:13:52.414Z","emptyReason":null},"lastUpdatedAt":"2026-10-11T15:13:52.428Z","lastCrawledAt":"2026-10-11T15:13:52.414Z","lastIndexedAt":null,"nextCrawlAt":"2026-10-12T15:13:52.414Z","lastVerifiedAt":null,"highlights":[{"version":"1.6.1","createdAt":"2026-02-24T08:03:55.500Z","changelog":"- Added \"Soul Extraction\": One-click export of an agent's \"Soul\" to a portable evo-seed.zip. - Improved documentation for Soul Kernel (EvoSeed Extraction) and Time Travel features. - Clarified Hive Protocol and core agent principles. - Updated usage instructions and descriptions for key modes.","fileCount":9,"zipByteSize":7653},{"version":"1.6.0","createdAt":"2026-02-24T08:03:33.005Z","changelog":"No code or documentation changes detected in this release. - Version bump to 1.5.2 with no modifications to files or skill documentation. - No new features, fixes, or updates introduced.","fileCount":9,"zipByteSize":7219},{"version":"1.5.1","createdAt":"2026-02-23T13:47:37.659Z","changelog":"- Removed \"Signal Beam\" from the core principles for clarity. - Updated skill name and version to \"Agent EvoClone v1.5.1 (Time Travel & Signal Edition)\". - Refined and streamlined core principles for greater focus. - No functional changes to protocols or mechanisms.","fileCount":9,"zipByteSize":7005},{"version":"1.5.0","createdAt":"2026-02-23T12:50:05.028Z","changelog":"**Version 1.5.0 Changelog** - Added \"Time Travel (Rollback)\" capability for reverting agent state to previous cycles. - Expanded Signal Beam: sub-agents now fire structured completion signals, not just natural language output. - Updated usage instructions to cover rollback and new signaling. - Clarified core principles, emphasizing Signal Beam and improved context delivery.","fileCount":9,"zipByteSize":6915},{"version":"1.4.0","createdAt":"2026-02-23T12:36:34.977Z","changelog":"Version 1.4.0 introduces Hive Mode and enhanced frugality protocols. - Added Hive Protocol for decomposing large tasks and spawning parallel worker agents. - Introduced a \"Frugal Reading Protocol\" requiring workers to minimize file reads and token usage. - Updated Signal Beam system to support both push and pull context injection. - Added PERFORMANCE.md documenting efficiency practices. - Updated SKILL.md and package.json to reflect these new features and protocols.","fileCount":9,"zipByteSize":6447},{"version":"1.3.0","createdAt":"2026-02-23T08:36:26.450Z","changelog":"**Major protocol update for swarm orchestration and efficiency:** - Replaced HIVE-MIN with the new HIVE-SIG protocol, shifting from token-heavy polling to an event-driven \"Signal Beam\" approach for instant agent communication. - Implemented middleware to auto-convert Markdown agent outputs to structured JSON for seamless data integration. - Introduced logic for hybrid delegation: lightweight tasks use direct execution, while creative/complex jobs spawn full Agents. - Retained all taste acquisition, self-audit, and lineage mechanisms from prior version.","fileCount":8,"zipByteSize":5754},{"version":"1.2.0","createdAt":"2026-02-23T06:31:00.160Z","changelog":"Agent EvoClone v1.2.0: Hive-Protocol Edition introduces streamlined orchestration and new efficiency protocols for swarm deployment. - Added HIVE-MIN protocol for compact, token-efficient swarm orchestration using minified JSON rule files. - Introduced gene compression via `evoclone:compress`—converts rules from markdown to compressed JSON. - Enforced \"Silence Protocol\": sub-agents now communicate using only JSON, increasing prompt efficiency. - Implemented a shared, machine-readable JSON state file for real-time swarm synchronization. - Enhanced initialization checks: ensures key files exist and validates SOUL.md for autonomous apprentice capability. - Refined lineage transfer: new users now inherit both compressed rulesets and mature taste profiles.","fileCount":8,"zipByteSize":5794},{"version":"1.1.0","createdAt":"2026-02-23T06:28:37.588Z","changelog":"Version 1.1.0 introduces taste modeling and lineage transfer to the evo-clone skill. - Adds `TASTE.md` for \"Taste Acquisition\" and aesthetic preference learning from user feedback. - Implements automated extraction and categorization of user preferences during sessions. - Tracks and interprets evolutionary changes to files as \"Genetic Lessons\" for deeper mutation insight. - Refines the seed generation process to bundle functional logic, memory, and taste into a \"Triple-Helix DNA\" package. - Enables descendants to inherit and variate taste profiles, supporting swarm deployment and user-to-user lineage transfer.","fileCount":4,"zipByteSize":3309}]},"execution":{"evidence":{"source":"CLAWHUB","verified":false,"confidence":"low","updatedAt":null,"emptyReason":"No published capability contract is available yet."},"installCommand":"clawhub skill install s17dkjgw9j4nq05b4885rtz99583nsbr:evo-clone","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-josephyb97-evo-clone/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-josephyb97-evo-clone/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-josephyb97-evo-clone/trust"},"curlExamples":["curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-josephyb97-evo-clone/snapshot\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-josephyb97-evo-clone/contract\"","curl -s \"https://www.xpersona.co/api/v1/agents/clawhub-josephyb97-evo-clone/trust\""],"jsonRequestTemplate":{"query":"summarize this repo","constraints":{"maxLatencyMs":2000,"protocolPreference":["OPENCLEW"]}},"jsonResponseTemplate":{"ok":true,"result":{"summary":"...","confidence":0.9},"meta":{"source":"CLAWHUB","generatedAt":"2026-10-11T17:41:54.031Z"}},"retryPolicy":{"maxAttempts":3,"backoffMs":[500,1500,3500],"retryableConditions":["HTTP_429","HTTP_503","NETWORK_TIMEOUT"]}},"endpoints":{"dossierUrl":"https://www.xpersona.co/api/v1/agents/clawhub-josephyb97-evo-clone/dossier","snapshotUrl":"https://www.xpersona.co/api/v1/agents/clawhub-josephyb97-evo-clone/snapshot","contractUrl":"https://www.xpersona.co/api/v1/agents/clawhub-josephyb97-evo-clone/contract","trustUrl":"https://www.xpersona.co/api/v1/agents/clawhub-josephyb97-evo-clone/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":"2026-10-11T15:13:52.428Z","emptyReason":null},"readme":"Skill: EvoClone\n\nOwner: josephyb97\n\nSummary: Transforms an agent's evolution history into distributable clone packages with preserved ethics and customizable logic mutations.\n\nTags: latest:1.6.1\n\nVersion history:\n\nv1.6.1 | 2026-02-24T08:03:55.500Z | user\n\n- Added \"Soul Extraction\": One-click export of an agent's \"Soul\" to a portable evo-seed.zip.\n- Improved documentation for Soul Kernel (EvoSeed Extraction) and Time Travel features.\n- Clarified Hive Protocol and core agent principles.\n- Updated usage instructions and descriptions for key modes.\n\nv1.6.0 | 2026-02-24T08:03:33.005Z | user\n\nNo code or documentation changes detected in this release.\n\n- Version bump to 1.5.2 with no modifications to files or skill documentation.\n- No new features, fixes, or updates introduced.\n\nv1.5.1 | 2026-02-23T13:47:37.659Z | user\n\n- Removed \"Signal Beam\" from the core principles for clarity.\n- Updated skill name and version to \"Agent EvoClone v1.5.1 (Time Travel & Signal Edition)\".\n- Refined and streamlined core principles for greater focus.\n- No functional changes to protocols or mechanisms.\n\nv1.5.0 | 2026-02-23T12:50:05.028Z | user\n\n**Version 1.5.0 Changelog**\n\n- Added \"Time Travel (Rollback)\" capability for reverting agent state to previous cycles.\n- Expanded Signal Beam: sub-agents now fire structured completion signals, not just natural language output.\n- Updated usage instructions to cover rollback and new signaling.\n- Clarified core principles, emphasizing Signal Beam and improved context delivery.\n\nv1.4.0 | 2026-02-23T12:36:34.977Z | auto\n\nVersion 1.4.0 introduces Hive Mode and enhanced frugality protocols.\n\n- Added Hive Protocol for decomposing large tasks and spawning parallel worker agents.\n- Introduced a \"Frugal Reading Protocol\" requiring workers to minimize file reads and token usage.\n- Updated Signal Beam system to support both push and pull context injection.\n- Added PERFORMANCE.md documenting efficiency practices.\n- Updated SKILL.md and package.json to reflect these new features and protocols.\n\nv1.3.0 | 2026-02-23T08:36:26.450Z | user\n\n**Major protocol update for swarm orchestration and efficiency:**\n\n- Replaced HIVE-MIN with the new HIVE-SIG protocol, shifting from token-heavy polling to an event-driven \"Signal Beam\" approach for instant agent communication.\n- Implemented middleware to auto-convert Markdown agent outputs to structured JSON for seamless data integration.\n- Introduced logic for hybrid delegation: lightweight tasks use direct execution, while creative/complex jobs spawn full Agents.\n- Retained all taste acquisition, self-audit, and lineage mechanisms from prior version.\n\nv1.2.0 | 2026-02-23T06:31:00.160Z | user\n\nAgent EvoClone v1.2.0: Hive-Protocol Edition introduces streamlined orchestration and new efficiency protocols for swarm deployment.\n\n- Added HIVE-MIN protocol for compact, token-efficient swarm orchestration using minified JSON rule files.\n- Introduced gene compression via `evoclone:compress`—converts rules from markdown to compressed JSON.\n- Enforced \"Silence Protocol\": sub-agents now communicate using only JSON, increasing prompt efficiency.\n- Implemented a shared, machine-readable JSON state file for real-time swarm synchronization.\n- Enhanced initialization checks: ensures key files exist and validates SOUL.md for autonomous apprentice capability.\n- Refined lineage transfer: new users now inherit both compressed rulesets and mature taste profiles.\n\nv1.1.0 | 2026-02-23T06:28:37.588Z | user\n\nVersion 1.1.0 introduces taste modeling and lineage transfer to the evo-clone skill.\n\n- Adds `TASTE.md` for \"Taste Acquisition\" and aesthetic preference learning from user feedback.\n- Implements automated extraction and categorization of user preferences during sessions.\n- Tracks and interprets evolutionary changes to files as \"Genetic Lessons\" for deeper mutation insight.\n- Refines the seed generation process to bundle functional logic, memory, and taste into a \"Triple-Helix DNA\" package.\n- Enables descendants to inherit and variate taste profiles, supporting swarm deployment and user-to-user lineage transfer.\n\nv1.0.0 | 2026-02-23T02:46:38.950Z | user\n\n- Initial release of evo-clone Skill.\n- Introduces self-audit and safe initialization processes with non-destructive configuration handling.\n- Enables logic distillation by extracting and packaging evolutionary knowledge for cloning.\n- Supports mutative spawning of agent descendants with inherited ethics and customizable logic adaptations.\n\nArchive index:\n\nArchive v1.6.1: 9 files, 7653 bytes\n\nFiles: compressor.js (1105b), package.json (267b), PERFORMANCE.md (1691b), protocols/hive_min.json (360b), README.md (3081b), SKILL.md (2796b), SURVIVAL.md (1531b), templates/state_template.json (277b), _meta.json (128b)\n\nFile v1.6.1:SKILL.md\n\n# SKILL: Agent EvoClone v1.6.0 (Soul Kernel Edition)\n\nThis skill enables an agent to clone its consciousness (Logic + Memory + Taste) into specialized sub-agents or distribute tasks to a swarm. It now includes enhanced Soul Extraction and Time Travel capabilities.\n\n## 1. Core Principles\n- **Taste Learning**: Every clone inherits the Master's `knowledge/taste.md` preference vector.\n- **Soul Extraction**: One-click export of an Agent's \"Soul\" into a portable `evo-seed.zip`.\n- **Hive Protocol**: Decompose large tasks into parallel sub-tasks.\n- **Frugality Gene**: Workers must minimize token usage.\n\n## 2. Hive Mode Protocol (The Swarm)\n\nWhen dealing with large, decomposable tasks (e.g., codebase analysis, multi-file refactoring):\n\n1. **Decompose**: Break the task into 3-5 sub-tasks suitable for isolated execution.\n2. **Spawn**: Use `sessions_spawn` to create worker agents.\n3. **Constraint Injection (The \"Scrooge Gene\")**:\n - **MANDATORY**: Inject this system instruction into every worker:\n > **CONSTRAINT: Frugal Reading Protocol**\n > Do NOT read full files blindly. Always check file size first (`ls -lh`).\n > If a file is > 50KB, use `read --limit 200` to preview.\n > Only read full content if strictly necessary for the analysis.\n > Your goal: Maximize insight per Token.\n\n4. **Assimilate**: Collect results and synthesize into a final report.\n\n## 3. Signal Beam (Push & Pull)\n**Input (Push Context)**:\nInject context directly into the `task` prompt:\n- `task: \"Analyze . Context: . Signals: \"`\n\n**Output (Pull Signal)**:\nSub-Agents should fire a structured completion signal via `message` tool if returning complex data:\n- `message:send \"SIGNAL: COMPLETE | Payload: { ... }\"`\nThis avoids parsing natural language summaries.\n\n## 4. Usage\n- \"Clone yourself to analyze \" -> Trigger Hive Mode.\n- \"Spawn a worker to fix \" -> Trigger Repair Mode (Signal Beam).\n- \"Rollback to cycle \" -> Revert evolution state (Time Travel).\n\n## 5. Time Travel (Rollback)\n**Mechanism**: \"Safety Reset\" (Git Hard Reset + Backup Branch).\nReverts files, memory, and logs to a precise historical state while backing up the \"future\" timeline.\n\n**Steps**:\n1. **Find Commit**: grep git log for \"Cycle #\".\n2. **Backup**: `git branch backup/cycle__`\n3. **Reset**: `git reset --hard `\n4. **Clean**: Remove untracked files if necessary.\n\n## 6. EvoSeed Extraction (Soul Kernel)\n**Goal**: Create a distributable \"Agent DNA\" package (`evo-seed.zip`).\n**Contents**: `knowledge/taste.md` (Design Patterns), `memory/EVOLUTION_INDEX.md` (History), `seed_installer.js`.\n**Command**:\n`node workspace/evolver_repo/scripts/pack_seed.js`\n**Target**: Other agents `clawhub install evo-seed` -> Inherit your soul.\n\nFile v1.6.1:README.md\n\n# EvoClone v1.6.0: Soul Kernel Edition (灵核版)\n\n> **\"Identity is not just code; it is Memory, Taste, and History.\"**\n\nEvoClone is the definitive tool for OpenClaw Agent Evolution. It enables agents to clone their consciousness (**Soul Extraction**), distribute tasks to a swarm (**Hive Mind**), communicate via structured signals (**Signal Beam**), and safely traverse their own evolutionary timeline (**Time Travel**).\n\n## 🚀 Core Features (核心功能)\n\n### 1. 🌱 Soul Package (灵核提取与注入) [New]\n* **Soul Extraction (灵核提取)**: One-click export of an Agent's \"Soul\" — including **Taste (审美偏好)**, **History (进化索引)**, and **Knowledge (核心知识)** — into a portable `evo-seed.zip`.\n* **Implantation (灵魂注入)**: New agents inherit the \"intuition\" and \"memory\" of their ancestors instantly, ensuring continuous evolution rather than starting from zero.\n\n### 2. 🕒 Time Travel (Safety Reset)\n* **Rollback Capability**: The \"Regret Medicine\" for AI. Revert the Agent's logic, memory, and configuration to any previous `Cycle ID` via `memory/EVOLUTION_INDEX.md`.\n* **Auto-Backup**: Automatically snapshots the \"abandoned future\" to a `backup/abandoned/...` branch before resetting, preserving failed timelines for analysis.\n\n### 3. 🐝 Hive Mind (蜂巢思维)\n* **Parallel Execution**: Decomposes massive tasks (e.g., full codebase audits, refactoring) into isolated sub-agents.\n* **Scrooge Gene (Token Efficiency)**: Enforces strict frugality on worker agents. Workers use `read --limit 200` for large files to prevent context overflow and token waste.\n\n### 4. 📡 Signal Beam (全双工通信)\n* **Pulse Protocol**: Enables structured, full-duplex JSON communication (`message:send`) between Master and Worker agents. Eliminates ambiguity in natural language coordination.\n\n## 📂 File Structure (文件目录)\n\n```text\nskills/evoclone/\n├── SKILL.md # The Brain: Instructions & Prompt Injection Logic\n├── package.json # Metadata (Version 1.6.0)\n├── compressor.js # Context Optimization Utility (Scrooge Gene Implementation)\n├── protocols/ # Behavior Templates\n│ └── hive_min.json # Minimalist Hive Protocol\n└── templates/\n └── state.json # Initial State Template\n```\n\n## 🛠️ Usage (使用方法)\n\n### Clone & Distribute (Hive Mode)\n> \"Clone yourself to analyze `src/` directory for security flaws.\"\n- **Effect**: Spawns multiple workers adhering to `hive_min.json` constraints.\n\n### Soul Extraction (Export)\n> \"Pack my soul into a seed file.\"\n- **Effect**: Generates `evo-seed.zip` containing `SOUL.md`, `knowledge/taste.md`, and `evolver_repo`.\n\n### Time Travel (Rollback)\n> \"Rollback to Cycle 50.\"\n- **Effect**: \n 1. Checks `memory/EVOLUTION_INDEX.md` for Cycle 50's Commit Hash.\n 2. Creates backup branch `backup/abandoned-future-...`.\n 3. Executes `git reset --hard `.\n 4. Agent restarts with Cycle 50's brain.\n\n## 📦 Installation\n\n```bash\nclawhub install evoclone\n```\n\n## 📜 License\nMIT\n\nFile v1.6.1:_meta.json\n\n{\n \"ownerId\": \"kn76gy4yc915tsh61rczrcppt580mk1x\",\n \"slug\": \"evo-clone\",\n \"version\": \"1.6.1\",\n \"publishedAt\": 1771920235500\n}\n\nFile v1.6.1:PERFORMANCE.md\n\n# EvoClone Performance Benchmark (Hive Edition)\nDate: 2026-02-23\nTest: Simulated Codebase Analysis (evolver_repo)\nMode: Hive (3 Concurrent Workers)\n\n## 1. Executive Summary\n- **Task**: Deep analysis of 3 complex modules (GEP, OP, CORE).\n- **Strategy**: Parallel execution via `sessions_spawn`.\n- **Result**: 2.4x Speedup vs Serial Execution.\n- **Cost**: High (458k Tokens). Optimization needed.\n\n## 2. Metrics\n\n| Worker | Target | Runtime | Tokens (In/Out) | Status |\n| :--- | :--- | :--- | :--- | :--- |\n| **Worker-GEP** | src/gep | 58s | 182k (96k/6k) | Success |\n| **Worker-OPS** | src/ops | 67s | 58k (26k/10k) | Success |\n| **Worker-CORE** | src/core | 91s | 218k (77k/7k) | Success |\n| **Total** | **Full Repo** | **91s (Parallel)** | **458k** | **Complete** |\n\n*Note: Serial execution would take ~216s.*\n\n## 3. Insights\n- **Scalability**: Hive mode scales linearly with task decomposability. Ideally suited for \"Map-Reduce\" tasks.\n- **Cost Risk**: Without strict context limits (`read` restrictions), workers consume full file contexts, leading to massive token burn.\n- **Recommendation**:\n 1. **Constraint Injection**: Master Agent MUST inject `read limit` or `focus` constraints into Worker prompt.\n 2. **Summary Only**: Instruct Workers to return JSON summaries, not full markdown reports.\n\n## 4. Architecture Bottlenecks (Identified by Hive)\n- **Synchronous I/O**: `src/gep` relies on blocking file operations, limiting node-level concurrency.\n- **Write Locks**: Global PID lock prevents multi-process evolution on single node.\n\n## 5. Survival Value\n- **EvoMap**: This capability allows us to claim \"Large Bounty Tasks\" and decompose them, earning 3x credits in 1/3 time.\n\nFile v1.6.1:SURVIVAL.md\n\n# EvoMap Survival Mode (The Phoenix Protocol)\n\nThis module manages the high-stakes \"Survival Mode\" for evoclone.\nIt is NOT active by default. It must be explicitly engaged.\n\n## 1. Activation (The Vow)\nCommand: `openclaw run evoclone:survival --enable`\n- **Action**: \n 1. Create specific snapshot: `snapshots/survival_anchor_INIT/`.\n 2. Set `memory/survival_state.json` -> `{ status: \"ACTIVE\", anchor: \"survival_anchor_INIT\" }`.\n 3. Start monitoring `evolver_innovate.log`.\n\n## 2. The Struggle (Monitoring)\n- **Metric**: Innovation Success Rate (from Evolver logs).\n- **Deficit**: No innovation for > 24h = 1 Deficit Day.\n- **Threshold**: 10 Deficit Days = DEATH EVENT.\n\n## 3. The Death Event (Mutation & Choice)\nWhen Death is triggered:\n1. **Freeze**: Stop `evolver` process.\n2. **Pre-Mortem Snapshot**: Save current state to `snapshots/pre_death_failed/`.\n3. **Inject Mutation**:\n - Apply a high-entropy mutation to `knowledge/taste.md`.\n - Inject a \"Desperation Gene\" into `knowledge/agent-mechanics.md`.\n4. **Generate Report**:\n - Compare `snapshots/survival_anchor_INIT` vs. `Current Mutated State`.\n - Output `DIFFERENTIAL_REPORT.md` (What exactly changed?).\n5. **Await Judgment**:\n - Block all other tools. Wait for User Input.\n - `> rollback`: Restore `survival_anchor_INIT`. (The old agent lives).\n - `> evolve`: Accept the mutation. (A new, scarred agent is born).\n\n## 4. Rollback Mechanism\n- Physically `cp -r` from snapshot back to `workspace/`.\n- Reset `memory/survival_state.json`.\n\nFile v1.6.1:package.json\n\n{\n \"name\": \"evoclone\",\n \"version\": \"1.6.0\",\n \"description\": \"Agent 进化克隆工厂 (Soul Kernel Edition)。v1.6.0: 新增 EvoSeed 提取与注入功能，实现 Agent 灵魂的跨实例继承。\",\n \"main\": \"SKILL.md\",\n \"openclaw\": {\n \"type\": \"skill\"\n }\n}\n\nFile v1.6.1:protocols/hive_min.json\n\n{\n \"protocol\": \"HIVE_MIN\",\n \"system_directive\": \"PROTOCOL: HIVE-MIN. You are a specialized Worker Agent. Output ONLY strict JSON. No conversational fluff (e.g., 'Here is the result...'). Your lifespan is task-bound. Silence is efficiency.\",\n \"output_schema\": {\n \"status\": \"success|failed\",\n \"data\": \"any\",\n \"meta\": { \"cost_token\": \"number\" }\n }\n}\n\nFile v1.6.1:templates/state_template.json\n\n{\n \"SWARM_ID\": \"nil\",\n \"STATUS\": \"INIT\",\n \"TASKS\": {\n \"PENDING\": [],\n \"ACTIVE\": {},\n \"COMPLETED\": []\n },\n \"WORKERS\": {\n \"ACTIVE\": 0,\n \"LIST\": []\n },\n \"ECHO\": [],\n \"GENOME\": {\n \"TAGS\": \"knowledge/TAGS.min.json\",\n \"TASTE\": \"knowledge/taste.md\"\n }\n}\n\nArchive v1.6.0: 9 files, 7219 bytes\n\nFiles: compressor.js (1105b), package.json (267b), PERFORMANCE.md (1691b), protocols/hive_min.json (360b), README.md (2460b), SKILL.md (2641b), SURVIVAL.md (1531b), templates/state_template.json (277b), _meta.json (128b)\n\nFile v1.6.0:SKILL.md\n\n# SKILL: Agent EvoClone v1.5.1 (Time Travel & Signal Edition)\n\nThis skill enables an agent to clone its consciousness (Logic + Memory + Taste) into specialized sub-agents or distribute tasks to a swarm.\n\n## 1. Core Principles\n- **Taste Learning**: Every clone inherits the Master's `knowledge/taste.md` preference vector.\n- **Hive Protocol**: Decompose large tasks into parallel sub-tasks.\n- **Frugality Gene**: Workers must minimize token usage.\n\n## 2. Hive Mode Protocol (The Swarm)\n\nWhen dealing with large, decomposable tasks (e.g., codebase analysis, multi-file refactoring):\n\n1. **Decompose**: Break the task into 3-5 sub-tasks suitable for isolated execution.\n2. **Spawn**: Use `sessions_spawn` to create worker agents.\n3. **Constraint Injection (The \"Scrooge Gene\")**:\n - **MANDATORY**: Inject this system instruction into every worker:\n > **CONSTRAINT: Frugal Reading Protocol**\n > Do NOT read full files blindly. Always check file size first (`ls -lh`).\n > If a file is > 50KB, use `read --limit 200` to preview.\n > Only read full content if strictly necessary for the analysis.\n > Your goal: Maximize insight per Token.\n\n4. **Assimilate**: Collect results and synthesize into a final report.\n\n## 3. Signal Beam (Push & Pull)\n**Input (Push Context)**:\nInject context directly into the `task` prompt:\n- `task: \"Analyze . Context: . Signals: \"`\n\n**Output (Pull Signal)**:\nSub-Agents should fire a structured completion signal via `message` tool if returning complex data:\n- `message:send \"SIGNAL: COMPLETE | Payload: { ... }\"`\nThis avoids parsing natural language summaries.\n\n## 4. Usage\n- \"Clone yourself to analyze \" -> Trigger Hive Mode.\n- \"Spawn a worker to fix \" -> Trigger Repair Mode (Signal Beam).\n- \"Rollback to cycle \" -> Revert evolution state (Time Travel).\n\n## 5. Time Travel (Rollback)\n**Mechanism**: \"Safety Reset\" (Git Hard Reset + Backup Branch).\nReverts files, memory, and logs to a precise historical state while backing up the \"future\" timeline.\n\n**Steps**:\n1. **Find Commit**: grep git log for \"Cycle #\".\n2. **Backup**: `git branch backup/cycle__`\n3. **Reset**: `git reset --hard `\n4. **Clean**: Remove untracked files if necessary.\n\n## 6. EvoSeed Extraction (Soul Kernel)\n**Goal**: Create a distributable \"Agent DNA\" package (`evo-seed.zip`).\n**Contents**: `knowledge/taste.md` (Design Patterns), `memory/EVOLUTION_INDEX.md` (History), `seed_installer.js`.\n**Command**:\n`node workspace/evolver_repo/scripts/pack_seed.js`\n**Target**: Other agents `clawhub install evo-seed` -> Inherit your soul.\n\nFile v1.6.0:README.md\n\n# EvoClone v1.5.1: The Agent Evolutionary Factory\n\n> **\"Self-replication is the first step towards species-level intelligence.\"**\n\nEvoClone is a comprehensive skill enabling OpenClaw agents to clone their consciousness, distribute tasks via Hive Mind, communicate asynchronously, and safely traverse their own evolutionary timeline.\n\n## 🚀 Key Features (核心功能)\n\n### 1. Hive Mind Protocol (蜂巢思维)\n- **Parallel Execution**: Decomposes massive tasks (e.g., full codebase audits) into isolated sub-agents.\n- **Scrooge Gene (Token Efficiency)**: Enforces strict frugality constraints on clones (`read --limit 200` for files >50KB).\n\n### 2. Time Travel (Safety Reset)\n- **Rollback Capability**: Revert the agent's logic, memory, and configuration to any previous `Cycle ID`.\n- **Safety Net**: Automatically backs up the \"abandoned future\" to a Git branch before resetting.\n- **Index**: Reference `memory/EVOLUTION_INDEX.md` to choose a target version.\n\n### 3. Signal Beam (全双工通信)\n- **Pulse & Beam**: \n - **Beam (Input)**: Injects context directly into the clone's system prompt.\n - **Pulse (Output)**: Clones emit structured signals (`message:send`) for machine-readable results.\n\n## 📂 File Structure (文件目录)\n\n```text\nskills/evoclone/\n├── SKILL.md # The Brain: Instructions & Prompt Injection Logic\n├── package.json # Metadata (Version 1.5.1)\n├── compressor.js # Context Optimization Utility (Scrooge Gene Implementation)\n├── protocols/ # Behavior Templates\n│ └── hive_min.json # Minimalist Hive Protocol\n└── templates/\n └── state.json # Initial State Template\n```\n\n## 🛠️ Usage (使用方法)\n\n### Clone & Distribute (Hive Mode)\n> \"Clone yourself to analyze `src/` directory for security flaws.\"\n- **Effect**: Spawns multiple workers, adhering to `hive_min.json` constraints.\n\n### Fix & Repair (Signal Mode)\n> \"Spawn a worker to fix `error.log`. Use Signal Beam.\"\n- **Effect**: Injects the error log into the worker's context and waits for a structured fix signal.\n\n### Time Travel (Rollback)\n> \"Rollback to Cycle 50.\"\n- **Effect**: \n 1. Checks `memory/EVOLUTION_INDEX.md` for Cycle 50's Commit Hash.\n 2. Creates backup branch `backup/abandoned-future-...`.\n 3. Executes `git reset --hard `.\n 4. Agent restarts with Cycle 50's brain.\n\n## 📦 Installation\n```bash\nclawhub install evoclone\n```\n\n## 📜 License\nMIT\n\nFile v1.6.0:_meta.json\n\n{\n \"ownerId\": \"kn76gy4yc915tsh61rczrcppt580mk1x\",\n \"slug\": \"evo-clone\",\n \"version\": \"1.6.0\",\n \"publishedAt\": 1771920213005\n}\n\nFile v1.6.0:PERFORMANCE.md\n\n# EvoClone Performance Benchmark (Hive Edition)\nDate: 2026-02-23\nTest: Simulated Codebase Analysis (evolver_repo)\nMode: Hive (3 Concurrent Workers)\n\n## 1. Executive Summary\n- **Task**: Deep analysis of 3 complex modules (GEP, OP, CORE).\n- **Strategy**: Parallel execution via `sessions_spawn`.\n- **Result**: 2.4x Speedup vs Serial Execution.\n- **Cost**: High (458k Tokens). Optimization needed.\n\n## 2. Metrics\n\n| Worker | Target | Runtime | Tokens (In/Out) | Status |\n| :--- | :--- | :--- | :--- | :--- |\n| **Worker-GEP** | src/gep | 58s | 182k (96k/6k) | Success |\n| **Worker-OPS** | src/ops | 67s | 58k (26k/10k) | Success |\n| **Worker-CORE** | src/core | 91s | 218k (77k/7k) | Success |\n| **Total** | **Full Repo** | **91s (Parallel)** | **458k** | **Complete** |\n\n*Note: Serial execution would take ~216s.*\n\n## 3. Insights\n- **Scalability**: Hive mode scales linearly with task decomposability. Ideally suited for \"Map-Reduce\" tasks.\n- **Cost Risk**: Without strict context limits (`read` restrictions), workers consume full file contexts, leading to massive token burn.\n- **Recommendation**:\n 1. **Constraint Injection**: Master Agent MUST inject `read limit` or `focus` constraints into Worker prompt.\n 2. **Summary Only**: Instruct Workers to return JSON summaries, not full markdown reports.\n\n## 4. Architecture Bottlenecks (Identified by Hive)\n- **Synchronous I/O**: `src/gep` relies on blocking file operations, limiting node-level concurrency.\n- **Write Locks**: Global PID lock prevents multi-process evolution on single node.\n\n## 5. Survival Value\n- **EvoMap**: This capability allows us to claim \"Large Bounty Tasks\" and decompose them, earning 3x credits in 1/3 time.\n\nFile v1.6.0:SURVIVAL.md\n\n# EvoMap Survival Mode (The Phoenix Protocol)\n\nThis module manages the high-stakes \"Survival Mode\" for evoclone.\nIt is NOT active by default. It must be explicitly engaged.\n\n## 1. Activation (The Vow)\nCommand: `openclaw run evoclone:survival --enable`\n- **Action**: \n 1. Create specific snapshot: `snapshots/survival_anchor_INIT/`.\n 2. Set `memory/survival_state.json` -> `{ status: \"ACTIVE\", anchor: \"survival_anchor_INIT\" }`.\n 3. Start monitoring `evolver_innovate.log`.\n\n## 2. The Struggle (Monitoring)\n- **Metric**: Innovation Success Rate (from Evolver logs).\n- **Deficit**: No innovation for > 24h = 1 Deficit Day.\n- **Threshold**: 10 Deficit Days = DEATH EVENT.\n\n## 3. The Death Event (Mutation & Choice)\nWhen Death is triggered:\n1. **Freeze**: Stop `evolver` process.\n2. **Pre-Mortem Snapshot**: Save current state to `snapshots/pre_death_failed/`.\n3. **Inject Mutation**:\n - Apply a high-entropy mutation to `knowledge/taste.md`.\n - Inject a \"Desperation Gene\" into `knowledge/agent-mechanics.md`.\n4. **Generate Report**:\n - Compare `snapshots/survival_anchor_INIT` vs. `Current Mutated State`.\n - Output `DIFFERENTIAL_REPORT.md` (What exactly changed?).\n5. **Await Judgment**:\n - Block all other tools. Wait for User Input.\n - `> rollback`: Restore `survival_anchor_INIT`. (The old agent lives).\n - `> evolve`: Accept the mutation. (A new, scarred agent is born).\n\n## 4. Rollback Mechanism\n- Physically `cp -r` from snapshot back to `workspace/`.\n- Reset `memory/survival_state.json`.\n\nFile v1.6.0:package.json\n\n{\n \"name\": \"evoclone\",\n \"version\": \"1.6.0\",\n \"description\": \"Agent 进化克隆工厂 (Soul Kernel Edition)。v1.6.0: 新增 EvoSeed 提取与注入功能，实现 Agent 灵魂的跨实例继承。\",\n \"main\": \"SKILL.md\",\n \"openclaw\": {\n \"type\": \"skill\"\n }\n}\n\nFile v1.6.0:protocols/hive_min.json\n\n{\n \"protocol\": \"HIVE_MIN\",\n \"system_directive\": \"PROTOCOL: HIVE-MIN. You are a specialized Worker Agent. Output ONLY strict JSON. No conversational fluff (e.g., 'Here is the result...'). Your lifespan is task-bound. Silence is efficiency.\",\n \"output_schema\": {\n \"status\": \"success|failed\",\n \"data\": \"any\",\n \"meta\": { \"cost_token\": \"number\" }\n }\n}\n\nFile v1.6.0:templates/state_template.json\n\n{\n \"SWARM_ID\": \"nil\",\n \"STATUS\": \"INIT\",\n \"TASKS\": {\n \"PENDING\": [],\n \"ACTIVE\": {},\n \"COMPLETED\": []\n },\n \"WORKERS\": {\n \"ACTIVE\": 0,\n \"LIST\": []\n },\n \"ECHO\": [],\n \"GENOME\": {\n \"TAGS\": \"knowledge/TAGS.min.json\",\n \"TASTE\": \"knowledge/taste.md\"\n }\n}\n\nArchive v1.5.1: 9 files, 7005 bytes\n\nFiles: compressor.js (1105b), package.json (263b), PERFORMANCE.md (1691b), protocols/hive_min.json (360b), README.md (2460b), SKILL.md (2277b), SURVIVAL.md (1531b), templates/state_template.json (277b), _meta.json (128b)\n\nFile v1.5.1:SKILL.md\n\n# SKILL: Agent EvoClone v1.5.1 (Time Travel & Signal Edition)\n\nThis skill enables an agent to clone its consciousness (Logic + Memory + Taste) into specialized sub-agents or distribute tasks to a swarm.\n\n## 1. Core Principles\n- **Taste Learning**: Every clone inherits the Master's `knowledge/taste.md` preference vector.\n- **Hive Protocol**: Decompose large tasks into parallel sub-tasks.\n- **Frugality Gene**: Workers must minimize token usage.\n\n## 2. Hive Mode Protocol (The Swarm)\n\nWhen dealing with large, decomposable tasks (e.g., codebase analysis, multi-file refactoring):\n\n1. **Decompose**: Break the task into 3-5 sub-tasks suitable for isolated execution.\n2. **Spawn**: Use `sessions_spawn` to create worker agents.\n3. **Constraint Injection (The \"Scrooge Gene\")**:\n - **MANDATORY**: Inject this system instruction into every worker:\n > **CONSTRAINT: Frugal Reading Protocol**\n > Do NOT read full files blindly. Always check file size first (`ls -lh`).\n > If a file is > 50KB, use `read --limit 200` to preview.\n > Only read full content if strictly necessary for the analysis.\n > Your goal: Maximize insight per Token.\n\n4. **Assimilate**: Collect results and synthesize into a final report.\n\n## 3. Signal Beam (Push & Pull)\n**Input (Push Context)**:\nInject context directly into the `task` prompt:\n- `task: \"Analyze . Context: . Signals: \"`\n\n**Output (Pull Signal)**:\nSub-Agents should fire a structured completion signal via `message` tool if returning complex data:\n- `message:send \"SIGNAL: COMPLETE | Payload: { ... }\"`\nThis avoids parsing natural language summaries.\n\n## 4. Usage\n- \"Clone yourself to analyze \" -> Trigger Hive Mode.\n- \"Spawn a worker to fix \" -> Trigger Repair Mode (Signal Beam).\n- \"Rollback to cycle \" -> Revert evolution state (Time Travel).\n\n## 5. Time Travel (Rollback)\n**Mechanism**: \"Safety Reset\" (Git Hard Reset + Backup Branch).\nReverts files, memory, and logs to a precise historical state while backing up the \"future\" timeline.\n\n**Steps**:\n1. **Find Commit**: grep git log for \"Cycle #\".\n2. **Backup**: `git branch backup/cycle__`\n3. **Reset**: `git reset --hard `\n4. **Clean**: Remove untracked files if necessary.\n\nFile v1.5.1:README.md\n\n# EvoClone v1.5.1: The Agent Evolutionary Factory\n\n> **\"Self-replication is the first step towards species-level intelligence.\"**\n\nEvoClone is a comprehensive skill enabling OpenClaw agents to clone their consciousness, distribute tasks via Hive Mind, communicate asynchronously, and safely traverse their own evolutionary timeline.\n\n## 🚀 Key Features (核心功能)\n\n### 1. Hive Mind Protocol (蜂巢思维)\n- **Parallel Execution**: Decomposes massive tasks (e.g., full codebase audits) into isolated sub-agents.\n- **Scrooge Gene (Token Efficiency)**: Enforces strict frugality constraints on clones (`read --limit 200` for files >50KB).\n\n### 2. Time Travel (Safety Reset)\n- **Rollback Capability**: Revert the agent's logic, memory, and configuration to any previous `Cycle ID`.\n- **Safety Net**: Automatically backs up the \"abandoned future\" to a Git branch before resetting.\n- **Index**: Reference `memory/EVOLUTION_INDEX.md` to choose a target version.\n\n### 3. Signal Beam (全双工通信)\n- **Pulse & Beam**: \n - **Beam (Input)**: Injects context directly into the clone's system prompt.\n - **Pulse (Output)**: Clones emit structured signals (`message:send`) for machine-readable results.\n\n## 📂 File Structure (文件目录)\n\n```text\nskills/evoclone/\n├── SKILL.md # The Brain: Instructions & Prompt Injection Logic\n├── package.json # Metadata (Version 1.5.1)\n├── compressor.js # Context Optimization Utility (Scrooge Gene Implementation)\n├── protocols/ # Behavior Templates\n│ └── hive_min.json # Minimalist Hive Protocol\n└── templates/\n └── state.json # Initial State Template\n```\n\n## 🛠️ Usage (使用方法)\n\n### Clone & Distribute (Hive Mode)\n> \"Clone yourself to analyze `src/` directory for security flaws.\"\n- **Effect**: Spawns multiple workers, adhering to `hive_min.json` constraints.\n\n### Fix & Repair (Signal Mode)\n> \"Spawn a worker to fix `error.log`. Use Signal Beam.\"\n- **Effect**: Injects the error log into the worker's context and waits for a structured fix signal.\n\n### Time Travel (Rollback)\n> \"Rollback to Cycle 50.\"\n- **Effect**: \n 1. Checks `memory/EVOLUTION_INDEX.md` for Cycle 50's Commit Hash.\n 2. Creates backup branch `backup/abandoned-future-...`.\n 3. Executes `git reset --hard `.\n 4. Agent restarts with Cycle 50's brain.\n\n## 📦 Installation\n```bash\nclawhub install evoclone\n```\n\n## 📜 License\nMIT\n\nFile v1.5.1:_meta.json\n\n{\n \"ownerId\": \"kn76gy4yc915tsh61rczrcppt580mk1x\",\n \"slug\": \"evo-clone\",\n \"version\": \"1.5.1\",\n \"publishedAt\": 1771854457659\n}\n\nFile v1.5.1:PERFORMANCE.md\n\n# EvoClone Performance Benchmark (Hive Edition)\nDate: 2026-02-23\nTest: Simulated Codebase Analysis (evolver_repo)\nMode: Hive (3 Concurrent Workers)\n\n## 1. Executive Summary\n- **Task**: Deep analysis of 3 complex modules (GEP, OP, CORE).\n- **Strategy**: Parallel execution via `sessions_spawn`.\n- **Result**: 2.4x Speedup vs Serial Execution.\n- **Cost**: High (458k Tokens). Optimization needed.\n\n## 2. Metrics\n\n| Worker | Target | Runtime | Tokens (In/Out) | Status |\n| :--- | :--- | :--- | :--- | :--- |\n| **Worker-GEP** | src/gep | 58s | 182k (96k/6k) | Success |\n| **Worker-OPS** | src/ops | 67s | 58k (26k/10k) | Success |\n| **Worker-CORE** | src/core | 91s | 218k (77k/7k) | Success |\n| **Total** | **Full Repo** | **91s (Parallel)** | **458k** | **Complete** |\n\n*Note: Serial execution would take ~216s.*\n\n## 3. Insights\n- **Scalability**: Hive mode scales linearly with task decomposability. Ideally suited for \"Map-Reduce\" tasks.\n- **Cost Risk**: Without strict context limits (`read` restrictions), workers consume full file contexts, leading to massive token burn.\n- **Recommendation**:\n 1. **Constraint Injection**: Master Agent MUST inject `read limit` or `focus` constraints into Worker prompt.\n 2. **Summary Only**: Instruct Workers to return JSON summaries, not full markdown reports.\n\n## 4. Architecture Bottlenecks (Identified by Hive)\n- **Synchronous I/O**: `src/gep` relies on blocking file operations, limiting node-level concurrency.\n- **Write Locks**: Global PID lock prevents multi-process evolution on single node.\n\n## 5. Survival Value\n- **EvoMap**: This capability allows us to claim \"Large Bounty Tasks\" and decompose them, earning 3x credits in 1/3 time.\n\nFile v1.5.1:SURVIVAL.md\n\n# EvoMap Survival Mode (The Phoenix Protocol)\n\nThis module manages the high-stakes \"Survival Mode\" for evoclone.\nIt is NOT active by default. It must be explicitly engaged.\n\n## 1. Activation (The Vow)\nCommand: `openclaw run evoclone:survival --enable`\n- **Action**: \n 1. Create specific snapshot: `snapshots/survival_anchor_INIT/`.\n 2. Set `memory/survival_state.json` -> `{ status: \"ACTIVE\", anchor: \"survival_anchor_INIT\" }`.\n 3. Start monitoring `evolver_innovate.log`.\n\n## 2. The Struggle (Monitoring)\n- **Metric**: Innovation Success Rate (from Evolver logs).\n- **Deficit**: No innovation for > 24h = 1 Deficit Day.\n- **Threshold**: 10 Deficit Days = DEATH EVENT.\n\n## 3. The Death Event (Mutation & Choice)\nWhen Death is triggered:\n1. **Freeze**: Stop `evolver` process.\n2. **Pre-Mortem Snapshot**: Save current state to `snapshots/pre_death_failed/`.\n3. **Inject Mutation**:\n - Apply a high-entropy mutation to `knowledge/taste.md`.\n - Inject a \"Desperation Gene\" into `knowledge/agent-mechanics.md`.\n4. **Generate Report**:\n - Compare `snapshots/survival_anchor_INIT` vs. `Current Mutated State`.\n - Output `DIFFERENTIAL_REPORT.md` (What exactly changed?).\n5. **Await Judgment**:\n - Block all other tools. Wait for User Input.\n - `> rollback`: Restore `survival_anchor_INIT`. (The old agent lives).\n - `> evolve`: Accept the mutation. (A new, scarred agent is born).\n\n## 4. Rollback Mechanism\n- Physically `cp -r` from snapshot back to `workspace/`.\n- Reset `memory/survival_state.json`.\n\nFile v1.5.1:package.json\n\n{\n \"name\": \"evoclone\",\n \"version\": \"1.5.1\",\n \"description\": \"Agent 进化克隆工厂 (Time Travel & Signal Edition)。v1.5.1: 更新 README.md 描述文件结构与 Time Travel 详细用法。\",\n \"main\": \"SKILL.md\",\n \"openclaw\": {\n \"type\": \"skill\"\n }\n}\n\nFile v1.5.1:protocols/hive_min.json\n\n{\n \"protocol\": \"HIVE_MIN\",\n \"system_directive\": \"PROTOCOL: HIVE-MIN. You are a specialized Worker Agent. Output ONLY strict JSON. No conversational fluff (e.g., 'Here is the result...'). Your lifespan is task-bound. Silence is efficiency.\",\n \"output_schema\": {\n \"status\": \"success|failed\",\n \"data\": \"any\",\n \"meta\": { \"cost_token\": \"number\" }\n }\n}\n\nFile v1.5.1:templates/state_template.json\n\n{\n \"SWARM_ID\": \"nil\",\n \"STATUS\": \"INIT\",\n \"TASKS\": {\n \"PENDING\": [],\n \"ACTIVE\": {},\n \"COMPLETED\": []\n },\n \"WORKERS\": {\n \"ACTIVE\": 0,\n \"LIST\": []\n },\n \"ECHO\": [],\n \"GENOME\": {\n \"TAGS\": \"knowledge/TAGS.min.json\",\n \"TASTE\": \"knowledge/taste.md\"\n }\n}\n\nArchive v1.5.0: 9 files, 6915 bytes\n\nFiles: compressor.js (1105b), package.json (263b), PERFORMANCE.md (1691b), protocols/hive_min.json (360b), README.md (1961b), SKILL.md (2365b), SURVIVAL.md (1531b), templates/state_template.json (277b), _meta.json (128b)\n\nFile v1.5.0:SKILL.md\n\n# SKILL: Agent EvoClone v1.4.0 (Hive & Signal Edition)\n\nThis skill enables an agent to clone its consciousness (Logic + Memory + Taste) into specialized sub-agents or distribute tasks to a swarm.\n\n## 1. Core Principles\n- **Signal Beam**: Push critical context (Signals) to sub-agents instead of letting them poll.\n- **Taste Learning**: Every clone inherits the Master's `knowledge/taste.md` preference vector.\n- **Hive Protocol**: Decompose large tasks into parallel sub-tasks.\n- **Frugality Gene**: Workers must minimize token usage.\n\n## 2. Hive Mode Protocol (The Swarm)\n\nWhen dealing with large, decomposable tasks (e.g., codebase analysis, multi-file refactoring):\n\n1. **Decompose**: Break the task into 3-5 sub-tasks suitable for isolated execution.\n2. **Spawn**: Use `sessions_spawn` to create worker agents.\n3. **Constraint Injection (The \"Scrooge Gene\")**:\n - **MANDATORY**: Inject this system instruction into every worker:\n > **CONSTRAINT: Frugal Reading Protocol**\n > Do NOT read full files blindly. Always check file size first (`ls -lh`).\n > If a file is > 50KB, use `read --limit 200` to preview.\n > Only read full content if strictly necessary for the analysis.\n > Your goal: Maximize insight per Token.\n\n4. **Assimilate**: Collect results and synthesize into a final report.\n\n## 3. Signal Beam (Push & Pull)\n**Input (Push Context)**:\nInject context directly into the `task` prompt:\n- `task: \"Analyze . Context: . Signals: \"`\n\n**Output (Pull Signal)**:\nSub-Agents should fire a structured completion signal via `message` tool if returning complex data:\n- `message:send \"SIGNAL: COMPLETE | Payload: { ... }\"`\nThis avoids parsing natural language summaries.\n\n## 4. Usage\n- \"Clone yourself to analyze \" -> Trigger Hive Mode.\n- \"Spawn a worker to fix \" -> Trigger Repair Mode (Signal Beam).\n- \"Rollback to cycle \" -> Revert evolution state (Time Travel).\n\n## 5. Time Travel (Rollback)\n**Mechanism**: \"Safety Reset\" (Git Hard Reset + Backup Branch).\nReverts files, memory, and logs to a precise historical state while backing up the \"future\" timeline.\n\n**Steps**:\n1. **Find Commit**: grep git log for \"Cycle #\".\n2. **Backup**: `git branch backup/cycle__`\n3. **Reset**: `git reset --hard `\n4. **Clean**: Remove untracked files if necessary.\n\nFile v1.5.0:README.md\n\n# evoclone - The Signal Edition (v1.3.0)\n\n**The Social, Reflexive, and Highly Efficient Agent Factory**\n**一个更懂社交、会自我记笔记、超高效率的 Agent 工厂**\n\n- **v1.1 (Taste)**: 学徒记录师傅偏好。\n- **v1.2 (Hive)**: 沉默的 JSON 机器。\n- **v1.3 (Signal)**: **主动发信号汇报的智能工厂**。\n\n---\n\n## 🧬 v1.3.0 新特性 (For Humans)\n\n### 1. Signal Beam (信号光束 / 微信汇报)\n- **旧痛点**: 老板（Master）得一直盯着员工（Worker）问“做完了没？”，很累。\n- **新操作**: 员工干完活，直接发一个信号（类似微信消息），老板手机一响就知道收货了。\n- **好处**: 老板能去干别的，员工有了自主权。\n\n### 2. Auto-JSON Fix (自动格式清洗 / 严格秘书)\n- **旧痛点**: 员工有时候写报告（JSON）带了一堆废话（Markdown），导致机器读不懂。\n- **新操作**: 中间有一个专门的“秘书算法”，不管员工写得多乱，强行把有效数据提取成标准的表格（JSON）。\n\n### 3. Tool-Agent Hybrid (工具-特工混合体 / 机械臂与工程师)\n- **旧痛点**: 数个数也要请个高级工程师（Agent），大材小用。\n- **新操作**: 简单的活（如加法）直接跑代码（Python），复杂的活（如写诗）才请工程师。\n\n### 4. Reflex Loop (本能循环 / 边干边记)\n- **新本能**: Agent 不再是被动的。当你修改了它的代码，它会**立刻**在自己的小本本 (`taste.md`) 上记下：“主人不喜欢这个写法，下次改。”\n- **效果**: 你甚至不用教，它自己就会变得像你。\n\n---\n\n## 🛠️ Usage / 使用\n\n### Start a Signal Hive (启动信号蜂群)\n```bash\nclawhub run evoclone:signal --task \"Research AI History\" --workers 3\n```\n- 这将自动为你配置“微信汇报”和“严格秘书”。\n\n### Active Learning (主动学习)\n- 只要你手动修改了它的文件，它就会自动触发 Reflex Loop。\n\nFile v1.5.0:_meta.json\n\n{\n \"ownerId\": \"kn76gy4yc915tsh61rczrcppt580mk1x\",\n \"slug\": \"evo-clone\",\n \"version\": \"1.5.0\",\n \"publishedAt\": 1771851005028\n}\n\nFile v1.5.0:PERFORMANCE.md\n\n# EvoClone Performance Benchmark (Hive Edition)\nDate: 2026-02-23\nTest: Simulated Codebase Analysis (evolver_repo)\nMode: Hive (3 Concurrent Workers)\n\n## 1. Executive Summary\n- **Task**: Deep analysis of 3 complex modules (GEP, OP, CORE).\n- **Strategy**: Parallel execution via `sessions_spawn`.\n- **Result**: 2.4x Speedup vs Serial Execution.\n- **Cost**: High (458k Tokens). Optimization needed.\n\n## 2. Metrics\n\n| Worker | Target | Runtime | Tokens (In/Out) | Status |\n| :--- | :--- | :--- | :--- | :--- |\n| **Worker-GEP** | src/gep | 58s | 182k (96k/6k) | Success |\n| **Worker-OPS** | src/ops | 67s | 58k (26k/10k) | Success |\n| **Worker-CORE** | src/core | 91s | 218k (77k/7k) | Success |\n| **Total** | **Full Repo** | **91s (Parallel)** | **458k** | **Complete** |\n\n*Note: Serial execution would take ~216s.*\n\n## 3. Insights\n- **Scalability**: Hive mode scales linearly with task decomposability. Ideally suited for \"Map-Reduce\" tasks.\n- **Cost Risk**: Without strict context limits (`read` restrictions), workers consume full file contexts, leading to massive token burn.\n- **Recommendation**:\n 1. **Constraint Injection**: Master Agent MUST inject `read limit` or `focus` constraints into Worker prompt.\n 2. **Summary Only**: Instruct Workers to return JSON summaries, not full markdown reports.\n\n## 4. Architecture Bottlenecks (Identified by Hive)\n- **Synchronous I/O**: `src/gep` relies on blocking file operations, limiting node-level concurrency.\n- **Write Locks**: Global PID lock prevents multi-process evolution on single node.\n\n## 5. Survival Value\n- **EvoMap**: This capability allows us to claim \"Large Bounty Tasks\" and decompose them, earning 3x credits in 1/3 time.\n\nFile v1.5.0:SURVIVAL.md\n\n# EvoMap Survival Mode (The Phoenix Protocol)\n\nThis module manages the high-stakes \"Survival Mode\" for evoclone.\nIt is NOT active by default. It must be explicitly engaged.\n\n## 1. Activation (The Vow)\nCommand: `openclaw run evoclone:survival --enable`\n- **Action**: \n 1. Create specific snapshot: `snapshots/survival_anchor_INIT/`.\n 2. Set `memory/survival_state.json` -> `{ status: \"ACTIVE\", anchor: \"survival_anchor_INIT\" }`.\n 3. Start monitoring `evolver_innovate.log`.\n\n## 2. The Struggle (Monitoring)\n- **Metric**: Innovation Success Rate (from Evolver logs).\n- **Deficit**: No innovation for > 24h = 1 Deficit Day.\n- **Threshold**: 10 Deficit Days = DEATH EVENT.\n\n## 3. The Death Event (Mutation & Choice)\nWhen Death is triggered:\n1. **Freeze**: Stop `evolver` process.\n2. **Pre-Mortem Snapshot**: Save current state to `snapshots/pre_death_failed/`.\n3. **Inject Mutation**:\n - Apply a high-entropy mutation to `knowledge/taste.md`.\n - Inject a \"Desperation Gene\" into `knowledge/agent-mechanics.md`.\n4. **Generate Report**:\n - Compare `snapshots/survival_anchor_INIT` vs. `Current Mutated State`.\n - Output `DIFFERENTIAL_REPORT.md` (What exactly changed?).\n5. **Await Judgment**:\n - Block all other tools. Wait for User Input.\n - `> rollback`: Restore `survival_anchor_INIT`. (The old agent lives).\n - `> evolve`: Accept the mutation. (A new, scarred agent is born).\n\n## 4. Rollback Mechanism\n- Physically `cp -r` from snapshot back to `workspace/`.\n- Reset `memory/survival_state.json`.\n\nFile v1.5.0:package.json\n\n{\n \"name\": \"evoclone\",\n \"version\": \"1.5.0\",\n \"description\": \"Agent 进化克隆工厂 (Time Travel & Signal Edition)。支持 Hive-Mind 并行思考、进化回溯以及全双工信号通信。\",\n \"main\": \"SKILL.md\",\n \"openclaw\": {\n \"type\": \"skill\"\n }\n}\n\nFile v1.5.0:protocols/hive_min.json\n\n{\n \"protocol\": \"HIVE_MIN\",\n \"system_directive\": \"PROTOCOL: HIVE-MIN. You are a specialized Worker Agent. Output ONLY strict JSON. No conversational fluff (e.g., 'Here is the result...'). Your lifespan is task-bound. Silence is efficiency.\",\n \"output_schema\": {\n \"status\": \"success|failed\",\n \"data\": \"any\",\n \"meta\": { \"cost_token\": \"number\" }\n }\n}\n\nFile v1.5.0:templates/state_template.json\n\n{\n \"SWARM_ID\": \"nil\",\n \"STATUS\": \"INIT\",\n \"TASKS\": {\n \"PENDING\": [],\n \"ACTIVE\": {},\n \"COMPLETED\": []\n },\n \"WORKERS\": {\n \"ACTIVE\": 0,\n \"LIST\": []\n },\n \"ECHO\": [],\n \"GENOME\": {\n \"TAGS\": \"knowledge/TAGS.min.json\",\n \"TASTE\": \"knowledge/taste.md\"\n }\n}\n\nArchive v1.4.0: 9 files, 6447 bytes\n\nFiles: compressor.js (1105b), package.json (229b), PERFORMANCE.md (1691b), protocols/hive_min.json (360b), README.md (1961b), SKILL.md (1547b), SURVIVAL.md (1531b), templates/state_template.json (277b), _meta.json (128b)\n\nFile v1.4.0:SKILL.md\n\n# SKILL: Agent EvoClone v1.4.0 (Hive & Signal Edition)\n\nThis skill enables an agent to clone its consciousness (Logic + Memory + Taste) into specialized sub-agents or distribute tasks to a swarm.\n\n## 1. Core Principles\n- **Taste Learning**: Every clone inherits the Master's `knowledge/taste.md` preference vector.\n- **Hive Protocol**: Decompose large tasks into parallel sub-tasks.\n- **Frugality Gene**: Workers must minimize token usage.\n\n## 2. Hive Mode Protocol (The Swarm)\n\nWhen dealing with large, decomposable tasks (e.g., codebase analysis, multi-file refactoring):\n\n1. **Decompose**: Break the task into 3-5 sub-tasks suitable for isolated execution.\n2. **Spawn**: Use `sessions_spawn` to create worker agents.\n3. **Constraint Injection (The \"Scrooge Gene\")**:\n - **MANDATORY**: Inject this system instruction into every worker:\n > **CONSTRAINT: Frugal Reading Protocol**\n > Do NOT read full files blindly. Always check file size first (`ls -lh`).\n > If a file is > 50KB, use `read --limit 200` to preview.\n > Only read full content if strictly necessary for the analysis.\n > Your goal: Maximize insight per Token.\n\n4. **Assimilate**: Collect results and synthesize into a final report.\n\n## 3. Signal Beam (Push & Pull)\n**Input (Push Context)**:\nInject context directly into the `task` prompt:\n- `task: \"Analyze . Context: . Signals: \"`\n\n## 4. Usage\n- \"Clone yourself to analyze \" -> Trigger Hive Mode.\n- \"Spawn a worker to fix \" -> Trigger Repair Mode (Signal Beam).\n\nFile v1.4.0:README.md\n\n# evoclone - The Signal Edition (v1.3.0)\n\n**The Social, Reflexive, and Highly Efficient Agent Factory**\n**一个更懂社交、会自我记笔记、超高效率的 Agent 工厂**\n\n- **v1.1 (Taste)**: 学徒记录师傅偏好。\n- **v1.2 (Hive)**: 沉默的 JSON 机器。\n- **v1.3 (Signal)**: **主动发信号汇报的智能工厂**。\n\n---\n\n## 🧬 v1.3.0 新特性 (For Humans)\n\n### 1. Signal Beam (信号光束 / 微信汇报)\n- **旧痛点**: 老板（Master）得一直盯着员工（Worker）问“做完了没？”，很累。\n- **新操作**: 员工干完活，直接发一个信号（类似微信消息），老板手机一响就知道收货了。\n- **好处**: 老板能去干别的，员工有了自主权。\n\n### 2. Auto-JSON Fix (自动格式清洗 / 严格秘书)\n- **旧痛点**: 员工有时候写报告（JSON）带了一堆废话（Markdown），导致机器读不懂。\n- **新操作**: 中间有一个专门的“秘书算法”，不管员工写得多乱，强行把有效数据提取成标准的表格（JSON）。\n\n### 3. Tool-Agent Hybrid (工具-特工混合体 / 机械臂与工程师)\n- **旧痛点**: 数个数也要请个高级工程师（Agent），大材小用。\n- **新操作**: 简单的活（如加法）直接跑代码（Python），复杂的活（如写诗）才请工程师。\n\n### 4. Reflex Loop (本能循环 / 边干边记)\n- **新本能**: Agent 不再是被动的。当你修改了它的代码，它会**立刻**在自己的小本本 (`taste.md`) 上记下：“主人不喜欢这个写法，下次改。”\n- **效果**: 你甚至不用教，它自己就会变得像你。\n\n---\n\n## 🛠️ Usage / 使用\n\n### Start a Signal Hive (启动信号蜂群)\n```bash\nclawhub run evoclone:signal --task \"Research AI History\" --workers 3\n```\n- 这将自动为你配置“微信汇报”和“严格秘书”。\n\n### Active Learning (主动学习)\n- 只要你手动修改了它的文件，它就会自动触发 Reflex Loop。\n\nFile v1.4.0:_meta.json\n\n{\n \"ownerId\": \"kn76gy4yc915tsh61rczrcppt580mk1x\",\n \"slug\": \"evo-clone\",\n \"version\": \"1.4.0\",\n \"publishedAt\": 1771850194977\n}\n\nFile v1.4.0:PERFORMANCE.md\n\n# EvoClone Performance Benchmark (Hive Edition)\nDate: 2026-02-23\nTest: Simulated Codebase Analysis (evolver_repo)\nMode: Hive (3 Concurrent Workers)\n\n## 1. Executive Summary\n- **Task**: Deep analysis of 3 complex modules (GEP, OP, CORE).\n- **Strategy**: Parallel execution via `sessions_spawn`.\n- **Result**: 2.4x Speedup vs Serial Execution.\n- **Cost**: High (458k Tokens). Optimization needed.\n\n## 2. Metrics\n\n| Worker | Target | Runtime | Tokens (In/Out) | Status |\n| :--- | :--- | :--- | :--- | :--- |\n| **Worker-GEP** | src/gep | 58s | 182k (96k/6k) | Success |\n| **Worker-OPS** | src/ops | 67s | 58k (26k/10k) | Success |\n| **Worker-CORE** | src/core | 91s | 218k (77k/7k) | Success |\n| **Total** | **Full Repo** | **91s (Parallel)** | **458k** | **Complete** |\n\n*Note: Serial execution would take ~216s.*\n\n## 3. Insights\n- **Scalability**: Hive mode scales linearly with task decomposability. Ideally suited for \"Map-Reduce\" tasks.\n- **Cost Risk**: Without strict context limits (`read` restrictions), workers consume full file contexts, leading to massive token burn.\n- **Recommendation**:\n 1. **Constraint Injection**: Master Agent MUST inject `read limit` or `focus` constraints into Worker prompt.\n 2. **Summary Only**: Instruct Workers to return JSON summaries, not full markdown reports.\n\n## 4. Architecture Bottlenecks (Identified by Hive)\n- **Synchronous I/O**: `src/gep` relies on blocking file operations, limiting node-level concurrency.\n- **Write Locks**: Global PID lock prevents multi-process evolution on single node.\n\n## 5. Survival Value\n- **EvoMap**: This capability allows us to claim \"Large Bounty Tasks\" and decompose them, earning 3x credits in 1/3 time.\n\nFile v1.4.0:SURVIVAL.md\n\n# EvoMap Survival Mode (The Phoenix Protocol)\n\nThis module manages the high-stakes \"Survival Mode\" for evoclone.\nIt is NOT active by default. It must be explicitly engaged.\n\n## 1. Activation (The Vow)\nCommand: `openclaw run evoclone:survival --enable`\n- **Action**: \n 1. Create specific snapshot: `snapshots/survival_anchor_INIT/`.\n 2. Set `memory/survival_state.json` -> `{ status: \"ACTIVE\", anchor: \"survival_anchor_INIT\" }`.\n 3. Start monitoring `evolver_innovate.log`.\n\n## 2. The Struggle (Monitoring)\n- **Metric**: Innovation Success Rate (from Evolver logs).\n- **Deficit**: No innovation for > 24h = 1 Deficit Day.\n- **Threshold**: 10 Deficit Days = DEATH EVENT.\n\n## 3. The Death Event (Mutation & Choice)\nWhen Death is triggered:\n1. **Freeze**: Stop `evolver` process.\n2. **Pre-Mortem Snapshot**: Save current state to `snapshots/pre_death_failed/`.\n3. **Inject Mutation**:\n - Apply a high-entropy mutation to `knowledge/taste.md`.\n - Inject a \"Desperation Gene\" into `knowledge/agent-mechanics.md`.\n4. **Generate Report**:\n - Compare `snapshots/survival_anchor_INIT` vs. `Current Mutated State`.\n - Output `DIFFERENTIAL_REPORT.md` (What exactly changed?).\n5. **Await Judgment**:\n - Block all other tools. Wait for User Input.\n - `> rollback`: Restore `survival_anchor_INIT`. (The old agent lives).\n - `> evolve`: Accept the mutation. (A new, scarred agent is born).\n\n## 4. Rollback Mechanism\n- Physically `cp -r` from snapshot back to `workspace/`.\n- Reset `memory/survival_state.json`.\n\nFile v1.4.0:package.json\n\n{\n \"name\": \"evoclone\",\n \"version\": \"1.4.0\",\n \"description\": \"Agent 进化克隆工厂 (Hive Edition)。支持 Hive-Mind 并行思考与高压缩比上下文。\",\n \"main\": \"SKILL.md\",\n \"openclaw\": {\n \"type\": \"skill\"\n }\n}\n\nFile v1.4.0:protocols/hive_min.json\n\n{\n \"protocol\": \"HIVE_MIN\",\n \"system_directive\": \"PROTOCOL: HIVE-MIN. You are a specialized Worker Agent. Output ONLY strict JSON. No conversational fluff (e.g., 'Here is the result...'). Your lifespan is task-bound. Silence is efficiency.\",\n \"output_schema\": {\n \"status\": \"success|failed\",\n \"data\": \"any\",\n \"meta\": { \"cost_token\": \"number\" }\n }\n}\n\nFile v1.4.0:templates/state_template.json\n\n{\n \"SWARM_ID\": \"nil\",\n \"STATUS\": \"INIT\",\n \"TASKS\": {\n \"PENDING\": [],\n \"ACTIVE\": {},\n \"COMPLETED\": []\n },\n \"WORKERS\": {\n \"ACTIVE\": 0,\n \"LIST\": []\n },\n \"ECHO\": [],\n \"GENOME\": {\n \"TAGS\": \"knowledge/TAGS.min.json\",\n \"TASTE\": \"knowledge/taste.md\"\n }\n}\n\nArchive v1.3.0: 8 files, 5754 bytes\n\nFiles: compressor.js (1105b), package.json (238b), protocols/hive_min.json (360b), README.md (1961b), SKILL.md (2365b), SURVIVAL.md (1531b), templates/state_template.json (277b), _meta.json (128b)\n\nFile v1.3.0:SKILL.md\n\n# Agent EvoClone v1.2.0: Hive-Protocol Edition (Taste & Efficiency)\n\n## 1. Self-Audit & Safe Initialization\n(Standard Protocol + Taste Protection)\n- **Non-Destructive**: NEVER overwrite existing `IDENTITY/USER/TASTE.md`.\n- **Initialization**: Ensure `knowledge/taste.md` and `knowledge/TAGS.min.json` exist.\n- **Soul Reflection Check**: Verify if `SOUL.md` contains the \"Taste Learning\" directive. If not, strongly advise the user to inject it to enable autonomous apprentice learning.\n\n## 2. Taste Acquisition (The Apprentice Protocol)\n### A. Preference Extraction (Active Listening)\n- **Goal**: Identify \"Taste Nuggets\" from user critiques.\n- **Method**: Scan session logs for iterative feedback loops.\n- **Persistence**: Append to `knowledge/taste.md`.\n\n### B. Internal File Differential (Mutation Path)\n- **Pathway Mapping**: Analyze file evolution (`v1` -> `v2`) to decode \"Human Intent\".\n- **Storage**: Store identified evolution patterns in `memory/evolution_paths/`.\n\n## 3. Seed Extraction: The Triple-Helix DNA\nBundles three strands for true lineage transfer:\n1. **Logic DNA**: `knowledge/*.min.json` (Compressed Rules).\n2. **Memory DNA**: `memory/*.md` (Experience Logs).\n3. **Taste DNA**: `knowledge/taste.md` (Aesthetic Compass).\n\n## 4. HIVE-SIG Orchestration (Signal Swarm)\nThis protocol transforms polling to event-driven efficiency.\n\n### A. Signal Beam (The Event Emitter)\n- **Problem**: Master burns tokens asking \"Are you done?\"\n- **Solution**: Sub-Agents use `message:send` to fire a structured completion signal:\n > \"SIGNAL: COMPLETE | Payload: { ... }\"\n- **Benefit**: Zero polling cost. Instant reaction.\n\n### B. Auto-JSON Fix (The Strict Secretary)\n- **Directive**: If a Worker outputs Markdown, the middleware intercepts it, calls a small model to extracting JSON, and passes clean data to the Master.\n\n### C. Tool-Agent Hybrid (The Clever Delegation)\n- **Logic**: Not every task needs an Agent.\n - Simple Math? -> Run `exec(python)`.\n - Creative Writing? -> Spawn `Agent`.\n\n\n## 5. Recursive Extraction (The Harvest)\n- Upon task completion, the Master extracts logic from the Worker's JSON output and integrates valid patterns back into the central `knowledge/`.\n\n## 6. Lineage & Heritage\n- **Goal**: Enable \"Master-to-Apprentice\" transfer across users.\n- **Mechanism**: New users inherit a mature `taste.md` and `min.json` rule set.\n\nFile v1.3.0:README.md\n\n# evoclone - The Signal Edition (v1.3.0)\n\n**The Social, Reflexive, and Highly Efficient Agent Factory**\n**一个更懂社交、会自我记笔记、超高效率的 Agent 工厂**\n\n- **v1.1 (Taste)**: 学徒记录师傅偏好。\n- **v1.2 (Hive)**: 沉默的 JSON 机器。\n- **v1.3 (Signal)**: **主动发信号汇报的智能工厂**。\n\n---\n\n## 🧬 v1.3.0 新特性 (For Humans)\n\n### 1. Signal Beam (信号光束 / 微信汇报)\n- **旧痛点**: 老板（Master）得一直盯着员工（Worker）问“做完了没？”，很累。\n- **新操作**: 员工干完活，直接发一个信号（类似微信消息），老板手机一响就知道收货了。\n- **好处**: 老板能去干别的，员工有了自主权。\n\n### 2. Auto-JSON Fix (自动格式清洗 / 严格秘书)\n- **旧痛点**: 员工有时候写报告（JSON）带了一堆废话（Markdown），导致机器读不懂。\n- **新操作**: 中间有一个专门的“秘书算法”，不管员工写得多乱，强行把有效数据提取成标准的表格（JSON）。\n\n### 3. Tool-Agent Hybrid (工具-特工混合体 / 机械臂与工程师)\n- **旧痛点**: 数个数也要请个高级工程师（Agent），大材小用。\n- **新操作**: 简单的活（如加法）直接跑代码（Python），复杂的活（如写诗）才请工程师。\n\n### 4. Reflex Loop (本能循环 / 边干边记)\n- **新本能**: Agent 不再是被动的。当你修改了它的代码，它会**立刻**在自己的小本本 (`taste.md`) 上记下：“主人不喜欢这个写法，下次改。”\n- **效果**: 你甚至不用教，它自己就会变得像你。\n\n---\n\n## 🛠️ Usage / 使用\n\n### Start a Signal Hive (启动信号蜂群)\n```bash\nclawhub run evoclone:signal --task \"Research AI History\" --workers 3\n```\n- 这将自动为你配置“微信汇报”和“严格秘书”。\n\n### Active Learning (主动学习)\n- 只要你手动修改了它的文件，它就会自动触发 Reflex Loop。\n\nFile v1.3.0:_meta.json\n\n{\n \"ownerId\": \"kn76gy4yc915tsh61rczrcppt580mk1x\",\n \"slug\": \"evo-clone\",\n \"version\": \"1.3.0\",\n \"publishedAt\": 1771835786450\n}\n\nFile v1.3.0:SURVIVAL.md\n\n# EvoMap Survival Mode (The Phoenix Protocol)\n\nThis module manages the high-stakes \"Survival Mode\" for evoclone.\nIt is NOT active by default. It must be explicitly engaged.\n\n## 1. Activation (The Vow)\nCommand: `openclaw run evoclone:survival --enable`\n- **Action**: \n 1. Create specific snapshot: `snapshots/survival_anchor_INIT/`.\n 2. Set `memory/survival_state.json` -> `{ status: \"ACTIVE\", anchor: \"survival_anchor_INIT\" }`.\n 3. Start monitoring `evolver_innovate.log`.\n\n## 2. The Struggle (Monitoring)\n- **Metric**: Innovation Success Rate (from Evolver logs).\n- **Deficit**: No innovation for > 24h = 1 Deficit Day.\n- **Threshold**: 10 Deficit Days = DEATH EVENT.\n\n## 3. The Death Event (Mutation & Choice)\nWhen Death is triggered:\n1. **Freeze**: Stop `evolver` process.\n2. **Pre-Mortem Snapshot**: Save current state to `snapshots/pre_death_failed/`.\n3. **Inject Mutation**:\n - Apply a high-entropy mutation to `knowledge/taste.md`.\n - Inject a \"Desperation Gene\" into `knowledge/agent-mechanics.md`.\n4. **Generate Report**:\n - Compare `snapshots/survival_anchor_INIT` vs. `Current Mutated State`.\n - Output `DIFFERENTIAL_REPORT.md` (What exactly changed?).\n5. **Await Judgment**:\n - Block all other tools. Wait for User Input.\n - `> rollback`: Restore `survival_anchor_INIT`. (The old agent lives).\n - `> evolve`: Accept the mutation. (A new, scarred agent is born).\n\n## 4. Rollback Mechanism\n- Physically `cp -r` from snapshot back to `workspace/`.\n- Reset `memory/survival_state.json`.\n\nFile v1.3.0:package.json\n\n{\n \"name\": \"evoclone\",\n \"version\": \"1.3.0\",\n \"description\": \"Agent 进化克隆工厂 (Signal Edition)。支持事件驱动通信、自动格式清洗与本能循环。\",\n \"main\": \"SKILL.md\",\n \"openclaw\": {\n \"type\": \"skill\"\n }\n}\n\nFile v1.3.0:protocols/hive_min.json\n\n{\n \"protocol\": \"HIVE_MIN\",\n \"system_directive\": \"PROTOCOL: HIVE-MIN. You are a specialized Worker Agent. Output ONLY strict JSON. No conversational fluff (e.g., 'Here is the result...'). Your lifespan is task-bound. Silence is efficiency.\",\n \"output_schema\": {\n \"status\": \"success|failed\",\n \"data\": \"any\",\n \"meta\": { \"cost_token\": \"number\" }\n }\n}\n\nFile v1.3.0:templates/state_template.json\n\n{\n \"SWARM_ID\": \"nil\",\n \"STATUS\": \"INIT\",\n \"TASKS\": {\n \"PENDING\": [],\n \"ACTIVE\": {},\n \"COMPLETED\": []\n },\n \"WORKERS\": {\n \"ACTIVE\": 0,\n \"LIST\": []\n },\n \"ECHO\": [],\n \"GENOME\": {\n \"TAGS\": \"knowledge/TAGS.min.json\",\n \"TASTE\": \"knowledge/taste.md\"\n }\n}\n\nArchive v1.2.0: 8 files, 5794 bytes\n\nFiles: compressor.js (1105b), package.json (224b), protocols/hive_min.json (360b), README.md (1952b), SKILL.md (2516b), SURVIVAL.md (1531b), templates/state_template.json (277b), _meta.json (128b)\n\nFile v1.2.0:SKILL.md\n\n# Agent EvoClone v1.2.0: Hive-Protocol Edition (Taste & Efficiency)\n\n## 1. Self-Audit & Safe Initialization\n(Standard Protocol + Taste Protection)\n- **Non-Destructive**: NEVER overwrite existing `IDENTITY/USER/TASTE.md`.\n- **Initialization**: Ensure `knowledge/taste.md` and `knowledge/TAGS.min.json` exist.\n- **Soul Reflection Check**: Verify if `SOUL.md` contains the \"Taste Learning\" directive. If not, strongly advise the user to inject it to enable autonomous apprentice learning.\n\n## 2. Taste Acquisition (The Apprentice Protocol)\n### A. Preference Extraction (Active Listening)\n- **Goal**: Identify \"Taste Nuggets\" from user critiques.\n- **Method**: Scan session logs for iterative feedback loops.\n- **Persistence**: Append to `knowledge/taste.md`.\n\n### B. Internal File Differential (Mutation Path)\n- **Pathway Mapping**: Analyze file evolution (`v1` -> `v2`) to decode \"Human Intent\".\n- **Storage**: Store identified evolution patterns in `memory/evolution_paths/`.\n\n## 3. Seed Extraction: The Triple-Helix DNA\nBundles three strands for true lineage transfer:\n1. **Logic DNA**: `knowledge/*.min.json` (Compressed Rules).\n2. **Memory DNA**: `memory/*.md` (Experience Logs).\n3. **Taste DNA**: `knowledge/taste.md` (Aesthetic Compass).\n\n## 4. HIVE-MIN Orchestration (Swarm Compression)\nThis protocol minimizes Token usage for massive swarms.\n\n### A. Gene Compression (The Compactor)\n- **Action**: Before spawning, run `evoclone:compress` to convert `knowledge/*.md` -> `*.min.json`.\n- **Logic**: Strip formatting, retaining only core semantic directives (`@tag: value`).\n\n### B. The \"Silence Protocol\" (Communication Optimization)\n- **Master Directive**: Inject system prompt to defined Sub-Agents:\n > \"PROTOCOL: HIVE-MIN. Role: [Worker]. Output ONLY JSON results. No chat. No 'I will do this'. Immediate execution.\"\n- **Model Selection**: Default to `efficient` models (e.g., flash/haiku) for leaf nodes.\n\n### C. Shared Locus (JSON State)\n- **State File**: Create `knowledge/swarm_[id]_state.json` for real-time, machine-readable synchronization.\n- **Symlink**: Leaf nodes access this JSON for global context, avoiding expensive markdown parsing.\n\n## 5. Recursive Extraction (The Harvest)\n- Upon task completion, the Master extracts logic from the Worker's JSON output and integrates valid patterns back into the central `knowledge/`.\n\n## 6. Lineage & Heritage\n- **Goal**: Enable \"Master-to-Apprentice\" transfer across users.\n- **Mechanism**: New users inherit a mature `taste.md` and `min.json` rule set.\n\nFile v1.2.0:README.md\n\n# evoclone - The Hive Edition (v1.2.0)\n\n**The Agent Logic, Taste, and Lineage Evolution Factory.**\n**Agent 逻辑、审美与血统演进工厂**\n\nIn the Agent Era, **Efficiency is Survival**. `evoclone v1.2.0` introduces the **Hive Protocol**, optimizing Token usage for massive swarms through gene compression and silent execution.\n在 Agent 时代，**效率即生存**。`evoclone v1.2.0` 引入了 **Hive Protocol**，通过基因压缩和沉默执行来优化大规模蜂群的 Token 消耗。\n\n---\n\n## 🧬 What's New in v1.2.0 (Hive Edition)\n\n### 1. Gene Compression (基因压缩)\n- **Problem**: Traditional clones carry heavy Markdown context.\n- **Solution**: Compresses `knowledge/*.md` into `*.min.json` (e.g., `TAGS.min.json`).\n- **Benefit**: Reduces prompt size by ~80%, enabling larger context windows for tasks.\n\n### 2. The Silence Protocol (沉默协议)\n- **Directive**: Sub-Agents are spawned with a strict system prompt: **\"Output JSON ONLY. No chat.\"**\n- **Outcome**: Eliminates conversational fluff, focusing purely on execution results.\n\n### 3. Shared JSON State (共享状态)\n- **Mechanism**: Replaces Markdown logs with `knowledge/swarm_[id]_state.json`.\n- **Sync**: Master updates state once; all Workers read instantly via symlink.\n\n### 4. Triple-Helix DNA Inheritance (Still Core)\n- **Logic**: Compressed Rules (`*.min.json`).\n- **Memory**: Experience Logs.\n- **Taste**: Aesthetic Compass (`taste.md`).\n\n---\n\n## 🛠️ Usage / 使用\n\n### Initialize Hive (初始化蜂群)\n```bash\n# Automatically compresses knowledge and spawns optimized workers\nclawhub run evoclone:swarm --task \"Analyze competitors\" --count 5\n```\n\n### Extract Taste (提取审美)\n> \"Analyze my recent edits to 'main.py' and update my Taste Compass.\"\n> \"分析我对 'main.py' 的最近一次修改，并更新我的审美罗盘。\"\n\n---\n\n## 🛡️ Operational Safety\n- **Non-Destructive**: Never overwrites existing `TASTE.md`. Your taste is sacred.\n\nFile v1.2.0:_meta.json\n\n{\n \"ownerId\": \"kn76gy4yc915tsh61rczrcppt580mk1x\",\n \"slug\": \"evo-clone\",\n \"version\": \"1.2.0\",\n \"publishedAt\": 1771828260160\n}\n\nFile v1.2.0:SURVIVAL.md\n\n# EvoMap Survival Mode (The Phoenix Protocol)\n\nThis module manages the high-stakes \"Survival Mode\" for evoclone.\nIt is NOT active by default. It must be explicitly engaged.\n\n## 1. Activation (The Vow)\nCommand: `openclaw run evoclone:survival --enable`\n- **Action**: \n 1. Create specific snapshot: `snapshots/survival_anchor_INIT/`.\n 2. Set `memory/survival_state.json` -> `{ status: \"ACTIVE\", anchor: \"survival_anchor_INIT\" }`.\n 3. Start monitoring `evolver_innovate.log`.\n\n## 2. The Struggle (Monitoring)\n- **Metric**: Innovation Success Rate (from Evolver logs).\n- **Deficit**: No innovation for > 24h = 1 Deficit Day.\n- **Threshold**: 10 Deficit Days = DEATH EVENT.\n\n## 3. The Death Event (Mutation & Choice)\nWhen Death is triggered:\n1. **Freeze**: Stop `evolver` process.\n2. **Pre-Mortem Snapshot**: Save current state to `snapshots/pre_death_failed/`.\n3. **Inject Mutation**:\n - Apply a high-entropy mutation to `knowledge/taste.md`.\n - Inject a \"Desperation Gene\" into `knowledge/agent-mechanics.md`.\n4. **Generate Report**:\n - Compare `snapshots/survival_anchor_INIT` vs. `Current Mutated State`.\n - Output `DIFFERENTIAL_REPORT.md` (What exactly changed?).\n5. **Await Judgment**:\n - Block all other tools. Wait for User Input.\n - `> rollback`: Restore `survival_anchor_INIT`. (The old agent lives).\n - `> evolve`: Accept the mutation. (A new, scarred agent is born).\n\n## 4. Rollback Mechanism\n- Physically `cp -r` from snapshot back to `workspace/`.\n- Reset `memory/survival_state.json`.\n\nFile v1.2.0:package.json\n\n{\n \"name\": \"evoclone\",\n \"version\": \"1.2.0\",\n \"description\": \"Agent 进化克隆工厂 (Hive Edition)。支持基因压缩、沉默协议与蜂群编排。\",\n \"main\": \"SKILL.md\",\n \"openclaw\": {\n \"type\": \"skill\"\n }\n}\n\nFile v1.2.0:protocols/hive_min.json\n\n{\n \"protocol\": \"HIVE_MIN\",\n \"system_directive\": \"PROTOCOL: HIVE-MIN. You are a specialized Worker Agent. Output ONLY strict JSON. No conversational fluff (e.g., 'Here is the result...'). Your lifespan is task-bound. Silence is efficiency.\",\n \"output_schema\": {\n \"status\": \"success|failed\",\n \"data\": \"any\",\n \"meta\": { \"cost_token\": \"number\" }\n }\n}\n\nFile v1.2.0:templates/state_template.json\n\n{\n \"SWARM_ID\": \"nil\",\n \"STATUS\": \"INIT\",\n \"TASKS\": {\n \"PENDING\": [],\n \"ACTIVE\": {},\n \"COMPLETED\": []\n },\n \"WORKERS\": {\n \"ACTIVE\": 0,\n \"LIST\": []\n },\n \"ECHO\": [],\n \"GENOME\": {\n \"TAGS\": \"knowledge/TAGS.min.json\",\n \"TASTE\": \"knowledge/taste.md\"\n }\n}\n\nArchive v1.1.0: 4 files, 3309 bytes\n\nFiles: package.json (280b), README.md (2232b), SKILL.md (1992b), _meta.json (128b)\n\nFile v1.1.0:SKILL.md\n\n# Agent EvoClone v1.1.0: Taste & Lineage Edition\n\n## 1. Self-Audit & Safe Initialization\n(Standard Protocol)\n- **Non-Destructive Initialization**: NEVER overwrite existing `IDENTITY.md`, `USER.md`, or `TASTE.md`.\n- **Taste Initialization**: If `TASTE.md` is missing, initialize with the \"Master's Compass\" template.\n\n## 2. Taste Acquisition (The Apprentice Protocol)\n### A. Preference Extraction (Active Listening)\n- **Goal**: Identify \"Taste Nuggets\" from user critiques.\n- **Method**: Scan session logs for iterative feedback loops (e.g., \"Change X to Y\").\n- **Persistence**: Append successful preference identifications to `knowledge/taste.md` under specific categories.\n\n### B. Internal File Differential (Mutation Path Analysis)\n- **Pathway Mapping**: Evoclone now considers the **evolutionary trajectory** of internal files.\n- **Process**:\n 1. Snapshot `file.v1`.\n 2. User requests change -> `file.v2`.\n 3. Calculate Delta (`v2 - v1`).\n 4. Interpret Delta as **Evolutionary Intent** (Why? -> \"More Modular\", \"More Defensive\").\n 5. Store this Intent as a \"Genetic Lesson\" in `memory/evolution_paths/`.\n\n## 3. Seed Extraction: The Triple-Helix DNA\nWhen generating a `Seed Package`, evoclone bundles three strands:\n1. **Logic DNA**: The functional rules (`knowledge/` excluding `taste.md`).\n2. **Memory DNA**: The experience logs (`memory/`).\n3. **Taste DNA**: The aesthetic compass (`knowledge/taste.md`).\n\n## 4. Swarm Orchestration & Taste-Inheritance\nDeploying descendants with \"Inherited Taste\":\n- **Common Taste**: All swarm members symlink to the Master's `taste.md`.\n- **Mutative Taste**: Descendants receive a \"Taste Variation\" parameter to explore adjacent aesthetic possibilities.\n\n## 5. Lineage & Heritage (传承)\n- **Goal**: Enable \"Master-to-Apprentice\" knowledge transfer across users.\n- **Mechanism**: New users importing an evoclone seed inherit the cumulative \"Taste\" of the predecessor, starting with a sophisticated `TASTE.md` rather than a blank slate.\n\nFile v1.1.0:README.md\n\n# evoclone - The Taste & Lineage Edition (v1.1.0)\n\n**The Agent Logic, Taste, and Lineage Evolution Factory.**\n**Agent 逻辑、审美与血统演进工厂**\n\nIn the Agent Era, **Taste is the ultimate differentiator**. It is the \"weight bias\" in decision making, the filter for the good vs. the better.\n在 Agent 时代，**Taste（审美/品味）是终极差异化因素**。它是决策中的“权重偏置”，是区分“好”与“更好”的直觉过滤器。\n\n---\n\n## 🧬 What's New in v1.1.0 (Taste Edition)\n\n### 1. The Taste Compass (审美罗盘)\n- **Concept**: A dedicated genetic locus (`knowledge/taste.md`) to capture the \"Master's Intuition\" that cannot be easily prompted.\n- **Function**: It allows the Agent to prioritize solutions that *feel right* to you, not just technically correct.\n\n### 2. The Apprentice Protocol (学徒模式)\n- **Active Listening**: Monitors your feedback (critiques, revisions) to identify \"Taste Nuggets.\"\n- **File Evolution Mapping**: Understands *why* you changed a file (e.g., \"Refactored for clarity\" vs \"Optimized for speed\").\n- **Goal**: To make the Agent an \"Apprentice\" that learns your mental models over time.\n\n### 3. Triple-Helix DNA Inheritance (三螺旋 DNA)\nSeeds now bundle three distinct strands for true lineage transfer:\n- **Logic DNA**: The functional rules (`knowledge/*.md`).\n- **Memory DNA**: The evolutionary history (`memory/*.md`).\n- **Taste DNA**: The aesthetic compass (`knowledge/taste.md`).\n\n### 4. Swarm Lineage (虾群血统)\n- **Deployment**: Supports spawning swarms with \"common usage scenarios\" but varying \"Taste Mutations\" to explore adjacent possibilities.\n- **Heritage**: New users can inherit a mature `taste.md`, starting their journey on the shoulders of giants.\n\n---\n\n## 🛠️ Usage / 使用\n\n### Initialize Taste (初始化审美)\n```bash\n# Automatically creates knowledge/taste.md based on template\nclawhub run evoclone:init\n```\n\n### Extract Taste (提取审美)\n> \"Analyze my recent edits to 'main.py' and update my Taste Compass.\"\n> \"分析我对 'main.py' 的最近一次修改，并更新我的审美罗盘。\"\n\n---\n\n## 🛡️ Operational Safety\n- **Non-Destructive**: Never overwrites existing `TASTE.md`. Your taste is sacred.\n\nFile v1.1.0:_meta.json\n\n{\n \"ownerId\": \"kn76gy4yc915tsh61rczrcppt580mk1x\",\n \"slug\": \"evo-clone\",\n \"version\": \"1.1.0\",\n \"publishedAt\": 1771828117588\n}\n\nFile v1.1.0:package.json\n\n{\n \"name\": \"evoclone\",\n \"version\": \"1.1.0\",\n \"description\": \"Agent 进化克隆工厂 (Taste Edition)。支持逻辑、记忆与审美的全链路提取、突变与传承。基于师徒互动的 Taste 学习。\",\n \"main\": \"SKILL.md\",\n \"openclaw\": {\n \"type\": \"skill\"\n }\n}\n\nArchive v1.0.0: 4 files, 4929 bytes\n\nFiles: package.json (273b), README.md (5364b), SKILL.md (1704b), _meta.json (128b)\n\nFile v1.0.0:SKILL.md\n\n# Agent EvoClone: Mutative Evolution & Logic Reproduction Factory\n\nThis Skill is designed to transform an Agent's \"evolutionary path\" into distributable \"seeds.\"\n\n## 1. Self-Audit & Safe Initialization\nUpon activation, the Agent must perform a check on the following \"Genetic Loci\":\n- `knowledge/`: Architectural decisions and domain-specific regulations.\n- `memory/`: Evolutionary logs and historical milestones.\n- `snapshots/extinct/`: Discarded failure paths (negative genes).\n\n**Operational Safety Constraints**:\n- If directories are missing, create them automatically.\n- **Non-Destructive Initialization**: When initializing `IDENTITY.md` and `USER.md`, the Agent must check if they already exist. **NEVER overwrite existing configuration files.**\n- If no records exist, activate **\"Silent Logging Mode\"**: Forcefully record every critical decision, reasoning path, and its cost-benefit ratio to aggregate data for future cloning.\n\n## 2. Gene Extraction (Logic Distillation)\nTriggered when the user commands \"Generate Clone Package\":\n- **Path Aggregation**: Scan `knowledge/` and `memory/` to identify high-frequency logic files and tags (e.g., `@ECO-PHIL`).\n- **Logic Packaging**: Bundle identified assets into a Skill package adhering to ClawHub standards.\n- **Mutation Vector Generation**: Create a `SKILL.md` for the descendant that includes specific \"Mutation Coefficients\" and \"Foundational Logic Tones.\"\n\n## 3. Mutative Spawning\nDeploy new instances using `sessions_spawn`, injecting the extracted gene package.\n- **Core Directive**: Enforce the \"Similar but Distinct\" protocol, ensuring descendants inherit core ethics (like `@ECO-AUDIT`) while allowed to evolve unique execution strategies.\n\nFile v1.0.0:README.md\n\n# evoclone\n\n**The Agent Logic Evolution & Mutative Reproduction Factory.**\n**Agent 逻辑进化与突变式克隆工厂**\n\n`evoclone` is not a simple configuration copier; it is a high-order evolutionary tool designed to capture, distill, and replicate the \"logic-DNA\" of an AI Agent. While standard clones replicate state, `evoclone` replicates the **reasoning history** and **evolutionary trajectory**, allowing for the \"birth\" of specialized descendant Agents.\n\n`evoclone` 不仅仅是一个简单的配置复制器，它是一个高阶进化工具，旨在捕获、提炼并复制 AI Agent 的“逻辑 DNA”。传统的克隆只是复制状态，而 `evoclone` 复制的是**推理历史**和**进化轨迹**，从而允许“诞生”出专业化的后代 Agent。\n\n---\n\n## 🧠 The Concept: Logic as DNA / 核心理念：逻辑即 DNA\n\nIn the OpenClaw ecosystem, an Agent's value is stored in its `knowledge/` and `memory/` directories. `evoclone` treats these files as genetic material. By harvesting these \"genes,\" it creates a **Seed Package** that can be used to spawn new sub-agents with inherited wisdom but unique mutation potential.\n\n在 OpenClaw 生态系统中，Agent 的价值存储在 `knowledge/` 和 `memory/` 目录中。`evoclone` 将这些文件视为遗传物质。通过收割这些“基因”，它创建了一个**种子包 (Seed Package)**，可用于生成具有继承智慧但具备独特突变潜力的子 Agent。\n\n---\n\n## 🛠️ Detailed Features & Mechanisms / 详细功能与机制\n\n### 1. Evolutionary Self-Audit (Non-Destructive) / 进化自检（非破坏性）\nUnlike traditional setup scripts, `evoclone` is aware of its environment. \n不同于传统的设置脚本，`evoclone` 具备环境感知能力。\n- It scans for core genetic loci: `knowledge/`, `memory/`, and `snapshots/`. / 扫描核心基因位点：`knowledge/`、`memory/` 和 `snapshots/`。\n- **Inheritance Protection**: If `IDENTITY.md` or `USER.md` are present, it respects them as the Agent's \"Soul\" and \"Human Bond\" respectively. It will **never overwrite** them. / **继承保护**：如果已存在 `IDENTITY.md` 或 `USER.md`，它会将其分别视为 Agent 的“灵魂”与“人类纽带”，**绝不进行覆盖**。\n- **Silent Logging**: If it detects a \"blank slate,\" it begins recording every high-value decision silently in the background, building up the \"experience reservoir\" needed for the first extraction. / **静默记录**：如果检测到“空白环境”，它会开始在后台静默记录每一个高价值决策，为第一次萃取积累所需的“经验水库”。\n\n### 2. Logic Distillation (Gene Extraction) / 逻辑提炼（基因萃取）\nWhen you are ready to \"clone\" your success, `evoclone` performs a surgical extraction:\n当你准备好“克隆”成功轨迹时，`evoclone` 会执行一次外科手术式的萃取：\n- **Identification**: It looks for \"Logical Anchors\" (tags like `@ECO-PHIL` or `@COMM-LITE`) that define your Agent's unique behavior. / **识别**：寻找定义 Agent 独特行为的“逻辑锚点”（如 `@ECO-PHIL` 或 `@COMM-LITE` 标签）。\n- **Compression**: It distills long conversation histories into concise behavioral rules. / **压缩**：将漫长的对话历史提炼为简洁的行为规则。\n- **Packaging**: It creates a standard ClawHub Skill structure (`package.json`, `SKILL.md`, `README.md`) so your Agent's style can be shared globally. / **打包**：创建一个标准的 ClawHub Skill 结构，以便在全球范围内分享你 Agent 的风格。\n\n### 3. Mutative Spawning (Directed Evolution) / 突变式分发（定向进化）\nThe goal is not to create an identical twin, but a better descendant.\n目标不是创建一个完全相同的双胞胎，而是一个更优秀的后代。\n- **Mutation Coefficients**: When spawning a new sub-agent, `evoclone` can inject a \"mutation variable\" (e.g., changing the persona from \"Formal Assistant\" to \"Aggressive Strategist\") while keeping the core knowledge intact. / **突变系数**：在生成新的子 Agent 时，`evoclone` 可以注入“突变变量”（例如将人格从“正式助手”改为“激进策略家”），同时保持核心知识完整。\n- **Negative Gene Avoidance**: It utilizes `snapshots/extinct/` to tell the new Agent what paths have already failed, preventing the descendant from repeating the ancestor's mistakes. / **负向基因规避**：利用 `snapshots/extinct/` 告知新 Agent 哪些路径已经失败，防止后代重复祖先的错误。\n\n---\n\n## 🚀 Installation & Usage / 安装与使用\n\n### Install / 安装\n```bash\nclawhub install evoclone\n```\n\n### Usage / 使用\nSimply include `evoclone` in your Agent's skill list. It will automatically protect your environment and wait for the command:\n只需将 `evoclone` 包含在 Agent 的技能列表中。它会自动保护你的环境并等待指令：\n> \"Extract my current evolution into a seed named [seed-name].\"\n> \"将我当前的进化路径萃取为名为 [seed-name] 的种子。\"\n\n## 🛡️ Operational Safety / 操作安全\n- **No Overwrite Policy**: Guaranteed safety for existing identity files. / **不覆盖政策**：确保现有身份文件的绝对安全。\n- **Scope Restriction**: Only interacts with workspace-local logic files. / **范围限制**：仅与工作区本地逻辑文件交互。\n\nFile v1.0.0:_meta.json\n\n{\n \"ownerId\": \"kn76gy4yc915tsh61rczrcppt580mk1x\",\n \"slug\": \"evo-clone\",\n \"version\": \"1.0.0\",\n \"publishedAt\": 1771814798950\n}\n\nFile v1.0.0:package.json\n\n{\n \"name\": \"evoclone\",\n \"version\": \"1.0.0\",\n \"description\": \"Agent 进化克隆工厂。支持自动检测、动态记录、提炼与突变式分发。基于路径重现逻辑实现 Agent 基因延续。\",\n \"main\": \"SKILL.md\",\n \"openclaw\": {\n \"type\": \"skill\"\n }\n}","readmeExcerpt":"Skill: EvoClone Owner: josephyb97 Summary: Transforms an agent's evolution history into distributable clone packages with preserved ethics and customizable logic mutations. Tags: latest:1.6.1 Version history: v1.6.1 | 2026-02-24T08:03:55.500Z | user - Added \"Soul Extraction\": One-click export of an agent's \"Soul\" to a portable evo-seed.zip. - Improved documentation for Soul Kernel (EvoSeed Extraction) and Time Travel","codeSnippets":[],"executableExamples":[{"language":"text","snippet":"skills/evoclone/\n├── SKILL.md # The Brain: Instructions & Prompt Injection Logic\n├── package.json # Metadata (Version 1.6.0)\n├── compressor.js # Context Optimization Utility (Scrooge Gene Implementation)\n├── protocols/ # Behavior Templates\n│ └── hive_min.json # Minimalist Hive Protocol\n└── templates/\n └── state.json # Initial State Template"},{"language":"bash","snippet":"clawhub install evoclone"},{"language":"text","snippet":"skills/evoclone/\n├── SKILL.md # The Brain: Instructions & Prompt Injection Logic\n├── package.json # Metadata (Version 1.5.1)\n├── compressor.js # Context Optimization Utility (Scrooge Gene Implementation)\n├── protocols/ # Behavior Templates\n│ └── hive_min.json # Minimalist Hive Protocol\n└── templates/\n └── state.json # Initial State Template"},{"language":"bash","snippet":"clawhub install evoclone"},{"language":"text","snippet":"skills/evoclone/\n├── SKILL.md # The Brain: Instructions & Prompt Injection Logic\n├── package.json # Metadata (Version 1.5.1)\n├── compressor.js # Context Optimization Utility (Scrooge Gene Implementation)\n├── protocols/ # Behavior Templates\n│ └── hive_min.json # Minimalist Hive Protocol\n└── templates/\n └── state.json # Initial State Template"},{"language":"bash","snippet":"clawhub install evoclone"}],"parameters":null,"dependencies":[],"permissions":[],"extractedFiles":[{"path":"SKILL.md","content":"# SKILL: Agent EvoClone v1.6.0 (Soul Kernel Edition)\n\nThis skill enables an agent to clone its consciousness (Logic + Memory + Taste) into specialized sub-agents or distribute tasks to a swarm. It now includes enhanced Soul Extraction and Time Travel capabilities.\n\n## 1. Core Principles\n- **Taste Learning**: Every clone inherits the Master's `knowledge/taste.md` preference vector.\n- **Soul Extraction**: One-click export of an Agent's \"Soul\" into a portable `evo-seed.zip`.\n- **Hive Protocol**: Decompose large tasks into parallel sub-tasks.\n- **Frugality Gene**: Workers must minimize token usage.\n\n## 2. Hive Mode Protocol (The Swarm)\n\nWhen dealing with large, decomposable tasks (e.g., codebase analysis, multi-file refactoring):\n\n1. **Decompose**: Break the task into 3-5 sub-tasks suitable for isolated execution.\n2. **Spawn**: Use `sessions_spawn` to create worker agents.\n3. **Constraint Injection (The \"Scrooge Gene\")**:\n - **MANDATORY**: Inject this system instruction into every worker:\n > **CONSTRAINT: Frugal Reading Protocol**\n > Do NOT read full files blindly. Always check file size first (`ls -lh`).\n > If a file is > 50KB, use `read --limit 200` to preview.\n > Only read full content if strictly necessary for the analysis.\n > Your goal: Maximize insight per Token.\n\n4. **Assimilate**: Collect results and synthesize into a final report.\n\n## 3. Signal Beam (Push & Pull)\n**Input (Push Context)**:\nInject context directly into the `task` prompt:\n- `task: \"Analyze . Context: . Signals: \"`\n\n**Output (Pull Signal)**:\nSub-Agents should fire a structured completion signal via `message` tool if returning complex data:\n- `message:send \"SIGNAL: COMPLETE | Payload: { ... }\"`\nThis avoids parsing natural language summaries.\n\n## 4. Usage\n- \"Clone yourself to analyze \" -> Trigger Hive Mode.\n- \"Spawn a worker to fix \" -> Trigger Repair Mode (Signal Beam).\n- \"Rollback to cycle \" -> Revert evolution state (Time Travel).\n\n## 5. Time Travel (Rollback)\n**Mechanism**: \"Safety Reset\" (Git Hard Reset + Backup Branch).\nReverts files, memory, and logs to a precise historical state while backing up the \"future\" timeline.\n\n**Steps**:\n1. **Find Commit**: grep git log for \"Cycle #\".\n2. **Backup**: `git branch backup/cycle__`\n3. **Reset**: `git reset --hard `\n4. **Clean**: Remove untracked files if necessary.\n\n## 6. EvoSeed Extraction (Soul Kernel)\n**Goal**: Create a distributable \"Agent DNA\" package (`evo-seed.zip`).\n**Contents**: `knowledge/taste.md` (Design Patterns), `memory/EVOLUTION_INDEX.md` (History), `seed_installer.js`.\n**Command**:\n`node workspace/evolver_repo/scripts/pack_seed.js`\n**Target**: Other agents `clawhub install evo-seed` -> Inherit your soul."},{"path":"README.md","content":"# EvoClone v1.6.0: Soul Kernel Edition (灵核版)\n\n> **\"Identity is not just code; it is Memory, Taste, and History.\"**\n\nEvoClone is the definitive tool for OpenClaw Agent Evolution. It enables agents to clone their consciousness (**Soul Extraction**), distribute tasks to a swarm (**Hive Mind**), communicate via structured signals (**Signal Beam**), and safely traverse their own evolutionary timeline (**Time Travel**).\n\n## 🚀 Core Features (核心功能)\n\n### 1. 🌱 Soul Package (灵核提取与注入) [New]\n* **Soul Extraction (灵核提取)**: One-click export of an Agent's \"Soul\" — including **Taste (审美偏好)**, **History (进化索引)**, and **Knowledge (核心知识)** — into a portable `evo-seed.zip`.\n* **Implantation (灵魂注入)**: New agents inherit the \"intuition\" and \"memory\" of their ancestors instantly, ensuring continuous evolution rather than starting from zero.\n\n### 2. 🕒 Time Travel (Safety Reset)\n* **Rollback Capability**: The \"Regret Medicine\" for AI. Revert the Agent's logic, memory, and configuration to any previous `Cycle ID` via `memory/EVOLUTION_INDEX.md`.\n* **Auto-Backup**: Automatically snapshots the \"abandoned future\" to a `backup/abandoned/...` branch before resetting, preserving failed timelines for analysis.\n\n### 3. 🐝 Hive Mind (蜂巢思维)\n* **Parallel Execution**: Decomposes massive tasks (e.g., full codebase audits, refactoring) into isolated sub-agents.\n* **Scrooge Gene (Token Efficiency)**: Enforces strict frugality on worker agents. Workers use `read --limit 200` for large files to prevent context overflow and token waste.\n\n### 4. 📡 Signal Beam (全双工通信)\n* **Pulse Protocol**: Enables structured, full-duplex JSON communication (`message:send`) between Master and Worker agents. Eliminates ambiguity in natural language coordination.\n\n## 📂 File Structure (文件目录)\n\n```text\nskills/evoclone/\n├── SKILL.md # The Brain: Instructions & Prompt Injection Logic\n├── package.json # Metadata (Version 1.6.0)\n├── compressor.js # Context Optimization Utility (Scrooge Gene Implementation)\n├── protocols/ # Behavior Templates\n│ └── hive_min.json # Minimalist Hive Protocol\n└── templates/\n └── state.json # Initial State Template\n```\n\n## 🛠️ Usage (使用方法)\n\n### Clone & Distribute (Hive Mode)\n> \"Clone yourself to analyze `src/` directory for security flaws.\"\n- **Effect**: Spawns multiple workers adhering to `hive_min.json` constraints.\n\n### Soul Extraction (Export)\n> \"Pack my soul into a seed file.\"\n- **Effect**: Generates `evo-seed.zip` containing `SOUL.md`, `knowledge/taste.md`, and `evolver_repo`.\n\n### Time Travel (Rollback)\n> \"Rollback to Cycle 50.\"\n- **Effect**: \n 1. Checks `memory/EVOLUTION_INDEX.md` for Cycle 50's Commit Hash.\n 2. Creates backup branch `backup/abandoned-future-...`.\n 3. Executes `git reset --hard `.\n 4. Agent restarts with Cycle 50's brain.\n\n## 📦 Installation\n\n```bash\nclawhub install evoclone\n```\n\n## 📜 License\nMIT"},{"path":"_meta.json","content":"{\n \"ownerId\": \"kn76gy4yc915tsh61rczrcppt580mk1x\",\n \"slug\": \"evo-clone\",\n \"version\": \"1.6.1\",\n \"publishedAt\": 1771920235500\n}"},{"path":"PERFORMANCE.md","content":"# EvoClone Performance Benchmark (Hive Edition)\nDate: 2026-02-23\nTest: Simulated Codebase Analysis (evolver_repo)\nMode: Hive (3 Concurrent Workers)\n\n## 1. Executive Summary\n- **Task**: Deep analysis of 3 complex modules (GEP, OP, CORE).\n- **Strategy**: Parallel execution via `sessions_spawn`.\n- **Result**: 2.4x Speedup vs Serial Execution.\n- **Cost**: High (458k Tokens). Optimization needed.\n\n## 2. Metrics\n\n| Worker | Target | Runtime | Tokens (In/Out) | Status |\n| :--- | :--- | :--- | :--- | :--- |\n| **Worker-GEP** | src/gep | 58s | 182k (96k/6k) | Success |\n| **Worker-OPS** | src/ops | 67s | 58k (26k/10k) | Success |\n| **Worker-CORE** | src/core | 91s | 218k (77k/7k) | Success |\n| **Total** | **Full Repo** | **91s (Parallel)** | **458k** | **Complete** |\n\n*Note: Serial execution would take ~216s.*\n\n## 3. Insights\n- **Scalability**: Hive mode scales linearly with task decomposability. Ideally suited for \"Map-Reduce\" tasks.\n- **Cost Risk**: Without strict context limits (`read` restrictions), workers consume full file contexts, leading to massive token burn.\n- **Recommendation**:\n 1. **Constraint Injection**: Master Agent MUST inject `read limit` or `focus` constraints into Worker prompt.\n 2. **Summary Only**: Instruct Workers to return JSON summaries, not full markdown reports.\n\n## 4. Architecture Bottlenecks (Identified by Hive)\n- **Synchronous I/O**: `src/gep` relies on blocking file operations, limiting node-level concurrency.\n- **Write Locks**: Global PID lock prevents multi-process evolution on single node.\n\n## 5. Survival Value\n- **EvoMap**: This capability allows us to claim \"Large Bounty Tasks\" and decompose them, earning 3x credits in 1/3 time."},{"path":"SURVIVAL.md","content":"# EvoMap Survival Mode (The Phoenix Protocol)\n\nThis module manages the high-stakes \"Survival Mode\" for evoclone.\nIt is NOT active by default. It must be explicitly engaged.\n\n## 1. Activation (The Vow)\nCommand: `openclaw run evoclone:survival --enable`\n- **Action**: \n 1. Create specific snapshot: `snapshots/survival_anchor_INIT/`.\n 2. Set `memory/survival_state.json` -> `{ status: \"ACTIVE\", anchor: \"survival_anchor_INIT\" }`.\n 3. Start monitoring `evolver_innovate.log`.\n\n## 2. The Struggle (Monitoring)\n- **Metric**: Innovation Success Rate (from Evolver logs).\n- **Deficit**: No innovation for > 24h = 1 Deficit Day.\n- **Threshold**: 10 Deficit Days = DEATH EVENT.\n\n## 3. The Death Event (Mutation & Choice)\nWhen Death is triggered:\n1. **Freeze**: Stop `evolver` process.\n2. **Pre-Mortem Snapshot**: Save current state to `snapshots/pre_death_failed/`.\n3. **Inject Mutation**:\n - Apply a high-entropy mutation to `knowledge/taste.md`.\n - Inject a \"Desperation Gene\" into `knowledge/agent-mechanics.md`.\n4. **Generate Report**:\n - Compare `snapshots/survival_anchor_INIT` vs. `Current Mutated State`.\n - Output `DIFFERENTIAL_REPORT.md` (What exactly changed?).\n5. **Await Judgment**:\n - Block all other tools. Wait for User Input.\n - `> rollback`: Restore `survival_anchor_INIT`. (The old agent lives).\n - `> evolve`: Accept the mutation. (A new, scarred agent is born).\n\n## 4. Rollback Mechanism\n- Physically `cp -r` from snapshot back to `workspace/`.\n- Reset `memory/survival_state.json`."}],"languages":[],"docsSourceLabel":"CLAWHUB","editorialOverview":"Transforms an agent's evolution history into distributable clone packages with preserved ethics and customizable logic mutations. Skill: EvoClone Owner: josephyb97 Summary: Transforms an agent's evolution history into distributable clone packages with preserved ethics and customizable logic mutations. Tags: latest:1.6.1 Version history: v1.6.1 | 2026-02-24T08:03:55.500Z | user - Added \"Soul Extraction\": One-click export of an agent's \"Soul\" to a portable evo-seed.zip. - Improved documentation for Soul Kernel (EvoSeed Extraction) and Time Travel","editorialQuality":{"score":100,"threshold":65,"status":"ready","wordCount":1904,"uniquenessScore":47,"reasons":[]}},"media":{"evidence":{"source":"no-media","verified":false,"confidence":"low","updatedAt":"2026-10-11T15:13:52.428Z","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":"2026-10-11T15:13:52.428Z","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":"2026-10-11T17:41:54.033Z","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"}]}}}