{"skill":{"slug":"feishu-agent-add","displayName":"feishuAgentAdd","summary":"Use this skill when users want to add a Feishu agent for OpenClaw, especially when they say things like “帮我增加一个名字叫xxx，用来做xxx的飞书agent”, want a guided prompt f...","description":"---\nname: feishu-agent-add\ndescription: Use this skill when users want to add a Feishu agent for OpenClaw, especially when they say things like “帮我增加一个名字叫xxx，用来做xxx的飞书agent”, want a guided prompt flow, or want a one-command way to generate the matching OpenClaw config.\nlicense: MIT\n---\n\n# feishu-agent-add\n\nThis skill is the conversational front end for the local script `scripts/add_feishu_agent.py`.\n\nThis project is designed for OpenClaw users, but the skill name intentionally stays short: `feishu-agent-add`.\n\n## When To Use\n\nUse this skill when the user wants to:\n\n- add a new Feishu-connected OpenClaw agent\n- avoid hand-editing `openclaw.json`\n- configure a new agent through a few follow-up questions\n- get a ready-to-run one-line command for advanced usage\n\n## Core Rule\n\nDo not hand-edit `openclaw.json` unless the user explicitly asks for manual fallback.\n\nPrefer running:\n\n```bash\npython3 scripts/add_feishu_agent.py ...\n```\n\nThe script is the execution core. This skill should mainly:\n\n1. understand the user's request\n2. ask only for missing required fields\n3. preview the plan\n4. run the script\n5. summarize the result and next steps\n\n## Required Inputs\n\nCollect these fields before execution:\n\n- `agent_name`\n- `purpose`\n- `agent_id`\n  - if missing, propose one derived from the name\n- `app_id`\n- `app_secret`\n\nThese can use defaults unless the user says otherwise:\n\n- `workspace_path`\n  - default: `~/.openclaw/workspace-{agent_id}`\n- `model`\n  - default: inherit from the current OpenClaw config\n- `enable_agent_to_agent`\n  - default: `true`\n- `workspace_mode`\n  - default: `auto`\n- `init_templates`\n  - default: `true`\n\n## Conversational Flow\n\n### 1. Parse what the user already gave\n\nFor a request like:\n\n> 帮我增加一个名字叫小红书运营，用来做内容选题和文案生成的飞书agent\n\nextract:\n\n- `agent_name = 小红书运营`\n- `purpose = 内容选题和文案生成`\n- `agent_id = xiaohongshu` or another short slug candidate\n\nIf `agent_id` is missing, propose one instead of asking an open-ended question.\n\n### 2. Ask the minimum follow-up questions\n\nOnly ask for the missing required fields. Prefer one compact message.\n\nTypical follow-up:\n\n- proposed `agent_id`\n- Feishu `App ID`\n- Feishu `App Secret`\n\nOnly ask about optional fields if the user indicates they care.\n\n### 3. Preview before execution\n\nBefore running the script, summarize:\n\n- agent name\n- agent id\n- purpose\n- workspace path\n- whether agent-to-agent collaboration will be enabled\n\n### 4. Run the script\n\nRun from the skill directory:\n\n```bash\npython3 scripts/add_feishu_agent.py \\\n  --agent-id <agent-id> \\\n  --agent-name \"<agent-name>\" \\\n  --purpose \"<purpose>\" \\\n  --app-id <app-id> \\\n  --app-secret <app-secret> \\\n  --json-output \\\n  --yes\n```\n\nAdd optional flags only when needed:\n\n- `--model <model>`\n- `--workspace-path <path>`\n- `--disable-agent-to-agent`\n- `--workspace-mode cli|mkdir|auto`\n- `--no-init-templates`\n- `--dry-run`\n\n## Advanced User Mode\n\nIf the user prefers a single terminal command, give them a ready-to-run example instead of a manual JSON recipe.\n\nUse this pattern:\n\n```bash\npython3 scripts/add_feishu_agent.py \\\n  --agent-id trader \\\n  --agent-name \"交易小助手\" \\\n  --purpose \"股票和 ETF 分析\" \\\n  --app-id cli_xxx \\\n  --app-secret secret_xxx \\\n  --yes\n```\n\n## Output Expectations\n\nAfter execution, summarize:\n\n- whether config was written successfully\n- the workspace path\n- whether starter files were initialized\n- that OpenClaw should be restarted\n- where to refine the agent identity, usually `SOUL.md`\n\nIf the script fails, report the concrete reason and do not improvise partial manual edits unless the user asks for that fallback.\n\n## Notes\n\n- The script already handles validation, backup, and config updates.\n- Prefer `--dry-run` first when the user asks for a preview.\n- If the user asks how to install or use this project, point them to `README.md`.\n","topics":["Feishu","飞书"],"tags":{"latest":"1.0.0"},"stats":{"comments":0,"downloads":653,"installsAllTime":25,"installsCurrent":0,"stars":0,"versions":1},"createdAt":1773553168625,"updatedAt":1778491917527},"latestVersion":{"version":"1.0.0","createdAt":1773553168625,"changelog":"Initial release of feishu-agent-add\n\n- Introduces a conversational skill for adding Feishu agents to OpenClaw without manual config editing.\n- Guides users through necessary input collection (agent name, purpose, app id/secret) with minimal questions.\n- Presents a summary preview before executing the setup script.\n- Runs a script to generate and update agent configuration, then summarizes results and next steps.\n- Supports both guided flows and ready-to-run terminal commands for advanced users.","license":"MIT-0"},"metadata":null,"owner":{"handle":"cyecho-io","userId":"s171h0p4vah5ddrf3shfbqftfx885aah","displayName":"cyecho-io","image":"https://avatars.githubusercontent.com/u/19159859?v=4"},"moderation":{"isSuspicious":false,"isMalwareBlocked":false,"verdict":"clean","reasonCodes":["review.llm_review"],"summary":"Review: review.llm_review","engineVersion":"v2.4.24","updatedAt":1780089893131}}