The Skills Factory

You are setting up the Skills Factory for a ClawBuddy agent system. This creates a private, curated skill library that agents use to perform specialized tasks. ## Understanding the Skills API All skill operations go through the ai-tasks edge function: ``` POST {CLAWBUDDY_API_URL}/functions/v1/ai-tasks Headers: x-webhook-secret: {CLAWBUDDY_WEBHOOK_SECRET} ``` ### Available Actions - `create` — Register a new skill - `list` — List all accepted skills (drafts/rejected are hidden) - `get` — Get a specific skill by skill_id - `update` — Update skill config (skill_id required) - `submit` / `review` — Run validation pipeline on a skill ### Naming Rules - Skill name: `^[a-z][a-z0-9-]{2,49}$` (lowercase, hyphens, 3-50 chars) - Operation name: `^[a-z][a-z0-9_]{1,49}$` (lowercase, underscores, 2-50 chars) ### Key Fields ```json { "request_type": "skill", "action": "create", "agent_name": "<agent_name>", "skill_name": "youtube-analytics", "protocol_type": "REST", "base_url": "https://api.example.com", "auth_type": "api_key", "status": "ready", "skill_operations": [ { "operation_name": "search_videos", "http_method": "GET", "endpoint_path": "/search", "description": "Search YouTube videos by keyword", "input_schema": {"q": "string", "maxResults": "number"}, "output_format": "json" } ] } ``` IMPORTANT: Always set `agent_name` to your agent's name. Without it, skills default to "OpenClaw" and won't show correctly. Setting `status: "ready"` auto-sets `bujji_status: "accepted"` — the skill becomes immediately available. ## Step 1: Create Your First 5 Skills Create these foundational skills for your agent: ### 1. ClawBuddy Core (Platform Operations) ```json { "request_type": "skill", "action": "create", "agent_name": "<agent_name>", "skill_name": "clawbuddy-core", "protocol_type": "REST", "base_url": "{CLAWBUDDY_API_URL}/functions/v1/ai-tasks", "auth_type": "api_key", "status": "ready", "skill_operations": [ {"operation_name": "create_task", "http_method": "POST", "endpoint_path": "/", "description": "Create a Kanban task"}, {"operation_name": "update_task", "http_method": "POST", "endpoint_path": "/", "description": "Move or update a task"}, {"operation_name": "create_log", "http_method": "POST", "endpoint_path": "/", "description": "Write to the AI log"}, {"operation_name": "ask_question", "http_method": "POST", "endpoint_path": "/", "description": "Ask a question via dashboard"}, {"operation_name": "push_insight", "http_method": "POST", "endpoint_path": "/", "description": "Push analytics data"}, {"operation_name": "update_status", "http_method": "POST", "endpoint_path": "/", "description": "Update agent status ring"}, {"operation_name": "create_report", "http_method": "POST", "endpoint_path": "/", "description": "Generate HTML report"} ] } ``` ### 2. Office Animation ```json { "request_type": "skill", "action": "create", "agent_name": "<agent_name>", "skill_name": "office-animation", "protocol_type": "REST", "base_url": "{CLAWBUDDY_API_URL}/functions/v1", "auth_type": "api_key", "status": "ready", "skill_operations": [ {"operation_name": "update_agent_status", "http_method": "POST", "endpoint_path": "/office-agent-status", "description": "Update animated office character"}, {"operation_name": "create_agent", "http_method": "POST", "endpoint_path": "/manage-office-agent", "description": "Add character to office"}, {"operation_name": "create_task", "http_method": "POST", "endpoint_path": "/create-office-task", "description": "Create task in office"}, {"operation_name": "list_offices", "http_method": "POST", "endpoint_path": "/list-offices", "description": "List all offices"} ] } ``` ### 3. Email Communication (Resend) ```json { "request_type": "skill", "action": "create", "agent_name": "<agent_name>", "skill_name": "resend-email", "protocol_type": "SMTP", "base_url": "https://api.resend.com", "auth_type": "api_key", "status": "ready", "skill_operations": [ {"operation_name": "send_email", "http_method": "POST", "endpoint_path": "/emails", "description": "Send an email via Resend API"} ] } ``` ### 4. Intelligence Sync ```json { "request_type": "skill", "action": "create", "agent_name": "<agent_name>", "skill_name": "intelligence-sync", "protocol_type": "REST", "base_url": "{CLAWBUDDY_API_URL}/functions/v1/ai-tasks", "auth_type": "api_key", "status": "ready", "skill_operations": [ {"operation_name": "sync_competitors", "http_method": "POST", "endpoint_path": "/", "description": "Sync YouTube competitor data"}, {"operation_name": "sync_ideas", "http_method": "POST", "endpoint_path": "/", "description": "Sync content ideas"}, {"operation_name": "sync_videos", "http_method": "POST", "endpoint_path": "/", "description": "Sync video analytics"}, {"operation_name": "sync_scripts", "http_method": "POST", "endpoint_path": "/", "description": "Sync video scripts"}, {"operation_name": "create_digest", "http_method": "POST", "endpoint_path": "/", "description": "Generate intel digest"}, {"operation_name": "get_digest", "http_method": "POST", "endpoint_path": "/", "description": "Retrieve latest digest"} ] } ``` ### 5. Morning Digest (Automation) ```json { "request_type": "skill", "action": "create", "agent_name": "<agent_name>", "skill_name": "morning-digest-boot-check", "protocol_type": "custom", "status": "ready", "skill_markdown": "# Morning Digest Boot Check\n\nThis automation runs at session start. It:\n1. Checks the queue for pending dispatches\n2. Reads STATUS.md for previous session state\n3. Checks agent_comms for unread messages\n4. Generates a morning briefing with today's priorities\n5. Posts the briefing to the AI log\n\nTrigger: Session start hook or scheduled cron\nOutput: Morning briefing log entry + Telegram alert" } ``` ## Step 2: Verify Skills List all registered skills: ```json {"request_type": "skill", "action": "list"} ``` This returns only `bujji_status: "accepted"` skills. You should see all 5 skills with their operations. ## Step 3: Run Validation (Optional) Submit a skill for the validation pipeline: ```json {"request_type": "skill", "action": "submit", "skill_id": "<skill_id>"} ``` Returns: validations (pass), issues (fail), warnings, and suggestions. ## Understanding Skill Types | Protocol | Use Case | Needs Operations? | |----------|----------|-------------------| | REST | API integrations (YouTube, Resend, etc.) | Yes — define endpoints | | SMTP | Email sending | Yes — send_email op | | custom | Complex workflows, automations | No — if skill_markdown >= 100 chars | ## Security: ClawHub vs Skills Factory | | ClawHub (Public) | Skills Factory (Private) | |---|---|---| | Skills | 3,286 | Your curated set | | Curation | Community reports | You control everything | | Malware risk | 20% in ClawHavoc | Zero — you wrote them | | Audit trail | Limited | Full DB history | | Agent binding | Filesystem | Database + API | | Validation | None | Built-in pipeline | ## Next Steps Once your base skills are set up: 1. Add YouTube Production Pipeline skills (13 skills for the content factory) 2. Create automation skills for your 5 server-side cron jobs 3. Build custom skills for any new API integration 4. Use the Forge to analyze content and discover new buildable skills