/install clawpage
clawpage
Share AI agent conversations as public web pages.
Supported Agents
| Agent | Profile |
|---|---|
| OpenClaw | references/platforms/openclaw.md |
| (unknown) | references/platforms/unknown.md — generic skill-based fallback |
| (new platform) | Add a file following references/platforms/TEMPLATE.md |
Core Workflow
1. Setup Check
- Detect agent type; load project dir + site URL using the agent profile
- If project not configured locally, ask the user:
- "Do you have an existing clawpage repo?"
- Yes → Existing Repo, New Environment
- No (Default) → First-Time Setup
2. Locate Session
- List sessions using agent profile discovery
- Show candidates → user confirms selection
3. Extract & Convert
Follow the Conversion section in the platform profile detected in Step 1.
Save the result to {projectDir}/chats/.tmp/{timestamp}.yaml.
4. Populate Metadata
The CLI auto-fills structural fields. The Skill's job is to fill in the human-facing metadata:
| Field | CLI default | Action |
|---|---|---|
date, sessionId, model, totalMessages, totalTokens, defaultShowProcess |
Auto-filled | Review only |
visibility |
private |
Update to public |
participants |
Generic role names (user, assistant) |
Ask user for display names → rename keys; if no name is provided, use Human / Agent for single-speaker or Human 1 / Human 2 … for multi-speaker |
title |
'Session Export' (generic) |
Skim generated YAML → suggest → confirm |
description |
(absent) | Write one-sentence summary → confirm |
channel |
(absent) | Ask user — set to platform name (e.g. discord) if applicable; omit otherwise |
cover |
(absent) | Skip (user adds custom OG image URL manually later) |
tags |
(absent) | Skip (user adds manually later) |
5. Redact
Review the generated YAML and replace sensitive information with [REDACTED]:
- API keys, tokens, passwords →
[REDACTED] - File paths with usernames (
/Users/xxx→~) - Email addresses, phone numbers →
[REDACTED] - Internal URLs and private IPs →
[REDACTED]
Do not silently remove content — always substitute [REDACTED] so the reader can see that content was present but withheld.
6. Confirm & Save
- Suggest filename:
{YYYYMMDD}-{topic}.yaml - Show preview → user confirms or modifies topic/filename
- Before moving the file, create a dedicated branch (see below — required even if the user does not publish yet):
cd {projectDir} git checkout main git pull origin main git checkout -b chat/{YYYYMMDD}-{topic} - Move:
{projectDir}/chats/.tmp/{timestamp}.yaml→{projectDir}/chats/{YYYYMMDD}-{topic}.yaml - Stage and commit immediately so the file is isolated on its own branch:
git add chats/{YYYYMMDD}-{topic}.yaml git commit -m "docs: add {topic}"
Why create a branch here? Saving on the default branch risks mixing unrelated changes into a future PR. Always commit each chat file on its own dedicated branch.
Optional: Publish
Push the branch created in step 6 and open a PR. See references/publish.md. Only proceed after explicit user request.
Edge Cases
- First-time project setup → references/setup.md
- Large or complex sessions (unknown platforms) → references/large-file.md
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install clawpage - 安装完成后,直接呼叫该 Skill 的名称或使用
/clawpage触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
Clawpage 是什么?
Share AI agent conversations as public web pages. Use when the user wants to share a conversation externally, export conversation history for documentation,... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 284 次。
如何安装 Clawpage?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install clawpage」即可一键安装,无需额外配置。
Clawpage 是免费的吗?
是的,Clawpage 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
Clawpage 支持哪些平台?
Clawpage 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 Clawpage?
由 yelo(@imyelo)开发并维护,当前版本 v1.0.1。