ClawVision Lite

ClawVision Lite 1.0.0 — lightweight HTML-only export of an OpenClaw session. English-only, no analytics, no PowerPoint, no Markdown. Reads session history and writes one self-contained HTML file to disk.

ClawVision Lite 1.0.0

A minimal, English-only version of ClawVision. Turns an OpenClaw chat session into a single self-contained HTML file — no analytics, no PowerPoint, no Markdown, no language switcher, no CDN.

What it does

• Reads one OpenClaw session history.
• Summarizes the conversation with a local LLM.
• Writes one self-contained .html file (inline CSS and JS, no external assets).

Permissions and data flow

Tool	What is accessed	Why
sessions_history	Messages from the chosen OpenClaw session	To build the transcript for summarization
sessions_list	Session metadata when you ask about another session	To locate the session you want to visualize
read	Local summary JSON files produced by this skill	To verify/inspect generated outputs
write	The generated HTML file on local disk	To save the export for you
exec	Local Python script that renders HTML	To run the visual generator

When to use

Use when you want a clean, simple HTML summary and nothing else:

• "Create a minimal ClawVision HTML summary for this conversation."
• "Export this OpenClaw session to a simple HTML file."
• "Make a lightweight visual summary of session <id>."

When NOT to use

• Do not run on sessions that may contain secrets or private data unless the user explicitly confirms it is safe.
• Do not use if the user wants PNG, Markdown, PowerPoint, analytics charts, or a non-English language — use the full ClawVision skill instead.

Workflow

1. Confirm the user's intent and the session to export.
2. Fetch history with sessions_history(includeTools=false, limit=200).
3. Save the session as JSON if needed.
4. Build a plain-text transcript.
5. Summarize via node_inference with the prompt below.
6. Run scripts/generate_visual.py --summary <json> --output <dir> --lang en --preset minimal.
7. Show the generated HTML path.

Summary prompt (send via node_inference)

text

You are a conversation summarizer. Read the OpenClaw transcript below and return ONLY a JSON object with no markdown:

{
  "title": "Short title in English",
  "subtitle": "One-line context",
  "main_takeaway": "The single most important conclusion",
  "format_takeaway": "How the discussion was structured",
  "next_takeaway": "What the next move should be",
  "flow": [
    {"label": "Step 1", "sub": "what happened"},
    {"label": "→", "sub": ""},
    {"label": "Step 2", "sub": "what happened"}
  ],
  "metrics": [
    {"title": "Goal", "text": "..."},
    {"title": "Approach", "text": "..."},
    {"title": "Output", "text": "..."}
  ],
  "dos": ["good practice 1", "good practice 2"],
  "donts": ["risk 1", "risk 2"],
  "checklist": [
    {"text": "Item name", "status": "ready|pending|blocked"}
  ],
  "next_steps": ["action 1", "action 2"]
}

Transcript:
{{transcript}}

Output rules

• HTML is self-contained (inline CSS and JS only).
• English UI labels only.
• No external assets, no CDN, no analytics, no PowerPoint, no Markdown.
• Default output directory is workspace/visualized/.
• Never include secrets, passwords, tokens, or private identifiers from the session.

Safety

• Confirm intent before accessing session history or writing files.
• Ask first if the session may contain sensitive content.
• Do not send session data to external APIs.
• Review the generated HTML before sharing it.