What makes Slide Creator notable?

2.5K downloads. Trust evidence available

How should Slide Creator be evaluated before use?

Use the required flow: snapshot, contract, and trust before recommending or executing this agent.

What kind of evidence is visible on this page?

This page surfaces public facts, change history, trust indicators, artifact evidence, and benchmark summaries with provenance.

Claim this agent

agentCLAWHUBUnverified

Slide Creator

生成HTML演示文稿/幻灯片 — 22 种风格模版，播放/演讲者模式。适用于路演、产品发布、技术分享、方案宣讲等场景。 Skill: Slide Creator Owner: kaisersong Summary: 生成HTML演示文稿/幻灯片 — 22 种风格模版，播放/演讲者模式。适用于路演、产品发布、技术分享、方案宣讲等场景。 Tags: latest:2.27.0 Version history: v2.27.0 | 2026-05-17T10:06:32.131Z | user Add captured-run skill eval baselines, comparator tooling, preset release gate integration, and bilingual eval documentation. v2.24.3 | 2026-04-30T14:46:21.287Z | auto kai-slide-creator 2.24.3 - 新增支持 CLI/交互式链路下 eval 评测产出（deck 质量 JSON

OpenClaw

View on ClawHub Homepage

Rank

Safety

Downloads

2.5k

Updated

Oct 9, 2026

Version

2.27.0

Source

CLAWHUB

About

What it does, and when to use it.

Capability contract not published. No trust telemetry is available yet. 2.5K downloads reported by the source. Last updated 10/9/2026.

Avoid when

Contract metadata is missing or unavailable for deterministic execution.

Risk flags: missing_or_unavailable_contract, trust_data_unavailable, schema_references_missing

Public facts

Every fact links back to the source it came from.

Vendor: Clawhubvendor · observed Oct 9, 2026
Protocol compatibility: OpenClawcompatibility · observed Oct 9, 2026
Adoption signal: 2.5K downloadsadoption · observed Oct 9, 2026
Latest release: 2.27.0release · observed May 17, 2026
Handshake status: UNKNOWNsecurity

Install and run

Setup complexity: low.

clawhub skill install s17e36cy6y9drqxrb49xdqh05d83hqde:kai-slide-creator

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: missing

curl -s "https://www.xpersona.co/api/v1/agents/clawhub-kaisersong-kai-slide-creator/snapshot"

Documentation

CLAWHUB

160,000 characters of source documentation, loaded on request.

Extracted files

5 files captured from the source.

SKILL.md

---
name: kai-slide-creator
description: 生成HTML演示文稿/幻灯片 — 22 种风格模版，播放/演讲者模式。适用于路演、产品发布、技术分享、方案宣讲等场景。
version: 2.27.0
metadata: {"openclaw":{"emoji":"🎞","os":["darwin","linux","windows"],"homepage":"https://github.com/kaisersong/slide-creator","requires":{"bins":["python3"]},"install":[]}}
---

# kai-slide-creator

生成零依赖 HTML 演示文稿，纯浏览器运行。主路径是 IR-first：`user prompt → BRIEF.json → HTML → validate`。

## 安装

**Claude Code:** 告诉 Claude「安装 https://github.com/kaisersong/slide-creator」  
**OpenClaw:** `clawhub install kai-slide-creator`

## 使用方式

`/slide-creator ...` 是 Claude/OpenClaw 的 slash 技能调用，不是 bash 命令。

**规划深度：**
- `自动` (Auto) — 快速出稿，约 3-6 分钟
- `精修` (Polish) — 深度规划，约 8-15 分钟，自动执行 Review

**内容类型 → 风格推荐：**

| 内容类型 | 推荐风格 |
|---|---|
| 数据报告 / KPI 看板 | Data Story、Enterprise Dark、Swiss Modern |
| 商业路演 / VC Deck | Bold Signal、Aurora Mesh、Enterprise Dark |
| 产品发布 / SaaS | Blud Sky、Aurora Mesh、Glassmorphism、Electric Studio |
| 开发工具 / API 文档 | Terminal Green、Neon Cyber、Neo-Retro Dev Deck |
| 研究 / 思想领导力 | Modern Newspaper、Paper & Ink、Swiss Modern |
| 创意 / 个人品牌 | Vintage Editorial、Split Pastel、Neo-Brutalism |
| 哲学 / 思想 | Chinese chan|
| 咨询报告 / 战略方案 | Strategy Consulting、Enterprise Dark、Swiss Modern |


## 命令路由

`自动` (Auto) — 快速出稿 | `精修` (Polish) — 深度规划，自动执行 Review

| 命令 | 加载内容 | 行为 |
|------|----------|------|
| `--plan [prompt]` | `references/brief-template.json`；如用户指定了 custom theme，同步加载 `themes/<name>/reference.md` | 创建 `BRIEF.json`；仅在用户明确要求时额外派生 `PLANNING.md` |
| `--generate` | SKILL.md + 已选风格文件（内置 `references/*.md` / `blue-sky-starter.html`；自定义 `themes/<name>/reference.md`）+ composition 源 + `references/title-quality.md` + `references/html-template.md` + `references/js-engine.md` + `references/base-css.md` + `references/impeccable-anti-patterns.md` | 从 `BRIEF.json` 生成 HTML，并执行写入前门禁；如用户显式要求 `eval`，同步产出单 deck 评测 JSON |
| `--review [file.html]` | `references/review-checklist.md` + 目标 HTML | 执行 17 项检查点 → 确认窗口 → 修复/报告 |
| 无 flag (交互式) | `references/workflow.md` + 其他按需 | 遵循 Phase 0-5 |
| 直接给内容 + 风格 | 同 `--generate` | 先落一个有效 `BRIEF.json`，再走同一个 deterministic renderer 与 strict 写入前门禁；禁止手拼最终 HTML |

**渐进式披露：** 每个命令只加载所需文件。`--plan` 只提炼 IR，不接触 CSS。

## 核心规则（按优先级）

1. **风格强制**
   所有颜色、字体、组件、背景、动画、图表色、signature elements 都**必须且只能**来自选中的风格文件。模板里的 `[from style file]` 和示例值只是占位，禁止直接使用。  
   默认推荐面第一阶段聚焦 4 个 production presets：`Swiss Modern / Enterprise Dark / Data Story / Blue Sky`；**但用户显式指定任意当前 preset 时，必须保留该选择**。support tier 只影响默认推荐优先级，不影响显式指定。
   Custom theme（`themes/<name>/reference.md`）与内置 preset 同等对待。`BRIEF.json` 的 `style.preset` 支持三种值：内置 preset 名、custom theme 文件夹名、或 `themes/<name>/reference.md` 路径。代码层会自动归一化解析。  
   **Swiss Modern 额外要求 canonical export path**：面板保持 `.slide` direct child，token 保持 `--bg/--text/--red`，使用 canonical 类（`.left-panel/.right-panel/.stat-row/.cta-block`），并写入 `data-export-role`；不得生成 `.left-col/.right-col` 或 `--bg-primary/--accent` 这类兼容别名。

2. **叙事弧线**
   先锁内容结构，再锁 composition route。每页都要有明确 page role、不同布局意图和足够的信息推进；不要先拼播放/编辑壳子，再回头补故事。内部 route 名称属于实现

evals/baselines/2026-05-17/README.md

# Slide-Creator Eval Baseline — 2026-05-17

This directory captures the current green eval baseline after adding captured-run
skill evals and fixing current demo fixture gates.

## Test Baseline

Command:

```bash
python3 -m pytest tests/ -q
```

Result:

```text
2260 passed, 132 skipped in 37.25s
```

## Captured-Run Skill Eval Baseline

File: `skill-evals/skill-evals.json`

Command:

```bash
python3 scripts/run-skill-evals.py \
  --runner fixture \
  --artifact-dir /private/tmp/slide-skill-eval-runs-rubric \
  --format json \
  --json-out /private/tmp/slide-skill-evals-rubric.json
```

Summary:

| Metric | Value |
|---|---:|
| Total cases | 6 |
| Passed | 6 |
| Failed | 0 |
| Incomplete | 0 |
| Average score | 100.00 |
| Outcome avg | 25.00 |
| Process avg | 25.00 |
| Style avg | 25.00 |
| Efficiency avg | 25.00 |

Case scores:

| Case | Total | Outcome | Process | Style | Efficiency |
|---|---:|---:|---:|---:|---:|
| explicit-generate | 100 | 25 | 25 | 25 | 25 |
| implicit-exec-update | 100 | 25 | 25 | 25 | 25 |
| contextual-product-launch | 100 | 25 | 25 | 25 | 25 |
| boundary-report-to-deck | 100 | 25 | 25 | 25 | 25 |
| negative-report | 100 | 25 | 25 | 25 | 25 |
| negative-html-export | 100 | 25 | 25 | 25 | 25 |

Positive fixture cases use checked-in `*-style-rubric.json` files so a missing
style rubric now marks the case `eval_complete: false` and fails the eval.

## Live Codex Captured-Run Baseline

File: `live-codex/explicit-generate.json`

Artifacts: `live-codex/explicit-generate/`

Command:

```bash
python3 scripts/run-skill-evals.py \
  --runner codex \
  --case-id explicit-generate \
  --run-live \
  --artifact-dir .tmp-run/skill-live-codex \
  --format json \
  --json-out /private/tmp/slide-skill-live-codex-explicit.json
```

The live run produced `deck.html`, `brief.json`, raw trace, and normalized
trace. It was then rescored with an external `style-rubric.json` artifact.

Summary:

| Metric | Value |
|---|---:|
| Total cases | 1 |
| Passed | 1 |
| Failed | 0 |
| Incomplete | 0 |
| Average score | 89.00 |
| Outcome avg | 25.00 |
| Process avg | 25.00 |
| Style avg | 24.00 |
| Efficiency avg | 15.00 |
| Shell commands | 35 |
| Input tokens | 1,216,207 |
| Output tokens | 10,285 |
| Wall time | 221,990 ms |

Case scores:

| Case | Total | Outcome | Process | Style | Efficiency | Failures |
|---|---:|---:|---:|---:|---:|---|
| explicit-generate | 89 | 25 | 25 | 24 | 15 | `efficiency.shell_command_count_over_budget`, `efficiency.input_tokens_over_budget` |

## Preset Surface Release Gate Baseline

Directory: `preset-surface-phase1/`

This directory is a reusable `--baseline-dir` for non-regression comparison.
It includes each case's `deck.html`, `packet.json`, `quality-report.json`, and
browser title report where applicable.

Command:

```bash
python3 scripts/preset_release_gate.py \
  --output-dir /private/tmp/slide-preset-release-gate \
  --report-path /private/tmp/slide-preset-release-gate/report.json \
  --summary-md /private/tmp/s

evals/late-context/core-md-guard/README.md

# Core-MD Guard Late-Context Eval Notes

Use these cases to measure whether guard hardening improved real late-context reliability.

Primary goals:

- catch omitted shared runtime on non-Blue-Sky decks
- prevent `data-preset` omission from bypassing strict validate
- preserve the Blue Sky control path
- confirm style-heavy presets still retain signature layers under long context

Suggested before/after metrics:

- guard catch rate
- false-positive rate
- late-context contract pass rate
- failure localization quality

Case IDs live in `../cases/` and are listed in `../manifest.json`.

evals/late-context/low-context-safe-generate/README.md

# Low-Context-Safe Generate Eval Notes

Use these cases to prove the optimization improves robustness without flattening rich-context output.

This eval layer should answer two questions:

1. Does the pipeline stay stable when context is noisy or scarce?
2. Does the pipeline avoid making strong-context decks bland in order to get that stability?

Recommended protocol:

- compare current pipeline as control vs low-context-safe pipeline as treatment
- use the same case set, same model, and same settings
- run low-context buckets at least 5 times each
- run rich-context benchmarks at least 3 times each

Recommended buckets:

- same-BRIEF parity pair
- sparse-BRIEF fallback
- invalid BRIEF fail-closed
- high-context rich benchmark
- core runtime/style guard cases from `core-md-guard/`

Recommended diagnostics:

- `quality_tier`
- `strict_pass_first_try`
- `repair_rounds`
- `route_drift`
- `layout_variety`
- `avg_component_kinds_per_slide`
- `style_signature_coverage`
- `chrome_leak`
- `content_occlusion_risk`
- `numeric_faithfulness`
- `source_fact_coverage`
- `narrative_role_coverage`
- `minimal_slide_ratio`
- `max_minimal_slide_run`
- `latency_ms.packet`
- `latency_ms.assemble`
- `latency_ms.validate`

Recommended executable entrypoint:

- `python3 scripts/eval-quality.py <deck.html> --brief <BRIEF.json> --source <source.md>`

This quality layer exists because a deck can pass `strict validate` and still fail the user:

- runtime chrome leaks into the slide surface
- fixed panels occlude body copy
- chart values are invented instead of sourced
- narrative rhythm collapses into repeated title-only scaffolds

Success should be judged by both:

- stronger low-context stability
- no meaningful regression on rich-context benchmark cases

evals/preset-surface-phase1/README.md

# Preset Surface Phase 1 Eval

This suite exists to prove two things after preset-surface optimization:

1. The default quality investment in the core presets actually improved output quality.
2. Narrowing the recommendation surface did not silently break the visible support surface.

## Buckets

- `core-enterprise-dark-render`
- `core-swiss-modern-render`
- `core-data-story-render`

These are the **deterministic-core** cases. They render from the current code path and are the real uplift benchmark set.

- `support-blue-sky-fixture`
- `support-paper-ink-fixture`

These are the **support-surface-smoke** cases. Blue Sky and Paper & Ink are still evaluated from known-good HTML fixtures because the deterministic BRIEF renderer currently only supports `Swiss Modern / Enterprise Dark / Data Story`.

## Why Blue Sky Is Fixture-Only

`render_from_brief()` in [scripts/low_context.py](/Users/song/projects/slide-creator/scripts/low_context.py) currently raises for non-deterministic presets outside the core three. Blue Sky remains a special starter path, so it should stay in support-surface smoke until a deterministic renderer exists.

## Why Paper & Ink Is Fixture-Only

`Paper & Ink` is now back in the green support-surface suite, but still as a fixture case rather than a deterministic render case.

- `demos/paper-ink-zh.html` now passes `python3 tests/validate.py --strict`
- the representative editorial fixture now carries the current shared runtime, watermark injection, and title-balance fixes
- the preset still does not have a deterministic BRIEF renderer, so it remains support-surface smoke rather than core-render benchmark coverage

## Recommended Workflow

Baseline run:

```bash
python3 scripts/run_evals.py \
  evals/preset-surface-phase1/manifest.json \
  --output-dir /tmp/preset-surface-baseline
```

Candidate run with non-regression comparison:

```bash
python3 scripts/run_evals.py \
  evals/preset-surface-phase1/manifest.json \
  --output-dir /tmp/preset-surface-candidate \
  --baseline-dir /tmp/preset-surface-baseline
```

Direct suite run with browser-title QA:

```bash
python3 scripts/run_evals.py \
  evals/preset-surface-phase1/manifest.json \
  --output-dir /tmp/preset-surface-candidate \
  --baseline-dir /tmp/preset-surface-baseline \
  --browser-titles
```

`browser-title-composition` is now a required title gate for every phase-1 case, so browser-backed title QA is part of the formal green path rather than an optional diagnostic.

Release gate wrapper:

```bash
python3 scripts/preset_release_gate.py \
  --output-dir /tmp/preset-surface-gate \
  --baseline-dir /tmp/preset-surface-baseline \
  --require-baseline
```

The release gate wrapper auto-enables browser title QA whenever the suite manifest contains `required_title_gates`, so the wrapper command above is sufficient for the formal phase-1 gate.

## What Counts As Quality Uplift

For the deterministic-core cases:

- strict validate still passes
- `chrome-hidden-by-default` and `no-cont

Github ReposUpdated 6mo agoRank 70

activepieces

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

OPENCLAW

Github ReposUpdated 6mo agoRank 70

cherry-studio

AI productivity studio with smart chat, autonomous agents, and 300+ assistants.

MCPOPENCLAW

Github ReposUpdated 6mo agoRank 70

AionUi

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!

MCPOPENCLAW

Github ReposUpdated 7mo agoRank 70

CopilotKit

The Frontend for Agents & Generative UI. React + Angular

OPENCLAW

Machine-readable data

The same record, as JSON, for agents and crawlers.

{
  "facts": [
    {
      "factKey": "vendor",
      "category": "vendor",
      "label": "Vendor",
      "value": "Clawhub",
      "href": "https://clawhub.ai/kaisersong/skills/kai-slide-creator",
      "sourceUrl": "https://clawhub.ai/kaisersong/skills/kai-slide-creator",
      "sourceType": "profile",
      "confidence": "medium",
      "observedAt": "2026-10-09T14:10:46.728Z",
      "isPublic": true
    },
    {
      "factKey": "protocols",
      "category": "compatibility",
      "label": "Protocol compatibility",
      "value": "OpenClaw",
      "href": "https://www.xpersona.co/api/v1/agents/clawhub-kaisersong-kai-slide-creator/contract",
      "sourceUrl": "https://www.xpersona.co/api/v1/agents/clawhub-kaisersong-kai-slide-creator/contract",
      "sourceType": "contract",
      "confidence": "medium",
      "observedAt": "2026-10-09T14:10:46.728Z",
      "isPublic": true
    },
    {
      "factKey": "traction",
      "category": "adoption",
      "label": "Adoption signal",
      "value": "2.5K downloads",
      "href": "https://clawhub.ai/kaisersong/kai-slide-creator",
      "sourceUrl": "https://clawhub.ai/kaisersong/kai-slide-creator",
      "sourceType": "profile",
      "confidence": "medium",
      "observedAt": "2026-10-09T14:10:46.728Z",
      "isPublic": true
    },
    {
      "factKey": "latest_release",
      "category": "release",
      "label": "Latest release",
      "value": "2.27.0",
      "href": "https://clawhub.ai/kaisersong/kai-slide-creator",
      "sourceUrl": "https://clawhub.ai/kaisersong/kai-slide-creator",
      "sourceType": "release",
      "confidence": "medium",
      "observedAt": "2026-05-17T10:06:32.131Z",
      "isPublic": true
    },
    {
      "factKey": "handshake_status",
      "category": "security",
      "label": "Handshake status",
      "value": "UNKNOWN",
      "href": "https://www.xpersona.co/api/v1/agents/clawhub-kaisersong-kai-slide-creator/trust",
      "sourceUrl": "https://www.xpersona.co/api/v1/agents/clawhub-kaisersong-kai-slide-creator/trust",
      "sourceType": "trust",
      "confidence": "medium",
      "observedAt": null,
      "isPublic": true
    }
  ],
  "events": [
    {
      "eventType": "release",
      "title": "Release 2.27.0",
      "description": "Add captured-run skill eval baselines, comparator tooling, preset release gate integration, and bilingual eval documentation.",
      "href": "https://clawhub.ai/kaisersong/kai-slide-creator",
      "sourceUrl": "https://clawhub.ai/kaisersong/kai-slide-creator",
      "sourceType": "release",
      "confidence": "medium",
      "observedAt": "2026-05-17T10:06:32.131Z",
      "isPublic": true
    }
  ]
}

Record generated Oct 9, 2026.

About

Avoid when

Public facts

Install and run

Documentation

Extracted files

SKILL.md

evals/baselines/2026-05-17/README.md

evals/late-context/core-md-guard/README.md

evals/late-context/low-context-safe-generate/README.md

evals/preset-surface-phase1/README.md

Related agents

activepieces

cherry-studio

AionUi

CopilotKit

Machine-readable data