--- name: clawdbot-monitor description: Monitor Clawdbot repo for updates and identify improvement opportunities for Geoffrey. Use when comparing Geoffrey to Clawdbot patterns, skills, or features. triggers: - "clawdbot monitor" - "check clawdbot" - "compare to clawdbot" - "sync with clawdbot" - "clawdbot updates" argument-hint: "[optional: focus area - skills|hooks|extensions|memory]" model: claude-opus-4-5-20251101 context: fork agent: Explore allowed-tools: - WebFetch - WebSearch - Read - Grep - Glob extended-thinking: true version: 1.0.0 --- # Clawdbot Monitor Monitor the [Clawdbot](https://github.com/clawdbot/clawdbot) repository and identify opportunities to improve Geoffrey. ## Clawdbot vs Geoffrey Quick Comparison | Aspect | Clawdbot | Geoffrey | |--------|----------|----------| | **Scope** | Multi-channel messaging hub (WhatsApp, Discord, etc.) | Claude Code plugin for personal AI | | **Skills** | 60+ (CLI wrappers, platform integrations) | 26 (specialized workflows) | | **Hooks** | Event-driven TypeScript handlers | hooks.json configuration | | **Memory** | Hybrid vector + BM25 with SQLite-vec | Obsidian vault + knowledge-manager | | **Releases** | Weekly (v2026.x.x latest) | Versioned plugin (v0.x.x) | ## Focus Areas When invoked with an argument, focus analysis on that area: - **skills** - Skill formats, metadata, categories, CLI patterns - **hooks** - Event types, handler patterns, automation - **extensions** - Plugin architecture, provider isolation - **memory** - Search patterns, embeddings, persistence Without an argument, perform full analysis across all areas. --- ## Phase 1: Fetch Clawdbot Current State ### 1.1 Core Documentation Fetch these from `github.com/clawdbot/clawdbot`: - `README.md` - Overall structure and philosophy - `CLAUDE.md` - System instructions (if exists) - Check releases page for version info ### 1.2 Skill Structure Explore the skills directory: - List all available skills - Sample key skills that overlap with Geoffrey's domains - Note skill metadata patterns (`requires.bins`, `requires.config`, install methods) - Check for new skills since last analysis ### 1.3 Hook System Examine hook architecture: - Event types (session, command, boot) - Handler patterns (TypeScript) - Compare to Geoffrey's hooks.json approach ### 1.4 Recent Changes - Check recent commits for significant updates - Look for new patterns or breaking changes - Note any announcements or migration guides --- ## Phase 2: Analyze Geoffrey Current State ### 2.1 Core Architecture Read these Geoffrey files: - `CLAUDE.md` - Founding principles and guidelines - `README.md` - Current capabilities - `.claude-plugin/plugin.json` - Version info ### 2.2 Skills Inventory ```bash # Glob for all skills skills/*/SKILL.md ``` Create inventory of current skills and their purposes. ### 2.3 Pattern Identification Identify Geoffrey's current patterns: - Hook system usage (hooks.json) - Knowledge storage approach (Obsidian vault) - Skill structure conventions (SKILL.md frontmatter) - Validation patterns --- ## Phase 3: Gap Analysis Compare Geoffrey against Clawdbot across these dimensions: ### 3.1 Skill Coverage - What Clawdbot skills do we lack equivalent skills for? - Which skills would provide highest value for Geoffrey's use case? - Are there redundant or outdated skills? ### 3.2 Skill Metadata System - `requires.bins` - CLI dependency declarations - `requires.config` - Configuration requirements - Install methods - How skills auto-install dependencies - Compare to Geoffrey's current skill structure ### 3.3 Hook System - Clawdbot: Event-driven TypeScript handlers - Geoffrey: hooks.json configuration - What event types are missing? - What patterns could we adopt? ### 3.4 Extension Architecture - Plugin isolation patterns - Multi-provider support (LLM providers, messaging channels) - How Clawdbot handles modularity ### 3.5 Memory System - Clawdbot: Hybrid vector + BM25 with SQLite-vec - Geoffrey: Obsidian vault + knowledge-manager - Embedding integration patterns - Search and retrieval approaches ### 3.6 New Features - Recent Clawdbot additions worth adopting - Experimental features to watch --- ## Phase 4: Generate Report Create a markdown report with this structure: ```markdown # Clawdbot Monitor Report **Geoffrey Version:** [from plugin.json] **Clawdbot Version Analyzed:** [from releases] **Analysis Date:** [today's date] ## Executive Summary [2-3 sentence summary of key findings] ## Clawdbot Recent Changes | Date | Change | Relevance to Geoffrey | |------|--------|----------------------| | ... | ... | ... | ## Features/Patterns to Adopt | Priority | Feature | Purpose | Complexity | Notes | |----------|---------|---------|------------|-------| | High | ... | ... | ... | ... | | Medium | ... | ... | ... | ... | | Low | ... | ... | ... | ... | ## Pattern Improvements ### 1. [Pattern Name] - **Clawdbot Implementation:** [how they do it] - **Geoffrey Current:** [how we do it / missing] - **Adoption Benefit:** [why adopt] - **Files to Modify:** [list] ### 2. [Pattern Name] ... ## Skill Coverage Comparison | Domain | Clawdbot Skill | Geoffrey Equivalent | Gap | |--------|---------------|--------------------| ----| | ... | ... | ... | ... | ## Recommended Actions ### Immediate (This Week) - [ ] Action 1 - [ ] Action 2 ### Short-Term (This Month) - [ ] Action 1 - [ ] Action 2 ### Long-Term (Explore) - [ ] Action 1 - [ ] Action 2 ``` --- ## Key Clawdbot Areas to Monitor Based on repo structure, these are the primary areas: | Area | Description | Geoffrey Equivalent | |------|-------------|-------------------| | skills/ | 60+ modular skills | skills/ (26) | | Hooks | Event-driven TypeScript | hooks.json | | Extensions | Plugin architecture | (none currently) | | Memory | Hybrid vector + BM25 | Obsidian vault | | Channels | WhatsApp, Discord, etc. | N/A (Claude Code only) | --- ## Key Differences from PAI Monitor | Aspect | PAI | Clawdbot | |--------|-----|----------| | Update frequency | Less frequent | Weekly releases | | Architecture | Packs + CLI | Daemon + extensions | | Skill count | 24 packs | 60+ skills | | Memory model | File-based | Vector DB (SQLite-vec) | | Target | Personal AI | Messaging hub | --- ## Execution Notes 1. **Start with Phase 1** - Always fetch fresh Clawdbot data, don't rely on cached knowledge 2. **Be thorough in Phase 2** - Accurate self-assessment is critical for gap analysis 3. **Prioritize in Phase 3** - Not all gaps need filling; focus on high-value opportunities for Geoffrey's use case 4. **Actionable Phase 4** - Every recommendation should have clear next steps 5. **Note Architecture Differences** - Clawdbot is a daemon, Geoffrey is a plugin; some patterns won't translate When focusing on a specific area, still provide brief context from other phases but concentrate analysis on the requested focus.