A production-grade skill for AI agents to interact with personal data through the Claw Agent Protocol.
The Claw Agent Protocol (CAP) Skill enables any AI agent to interact with a user's personal data in a structured, secure, and efficient manner. This skill provides a comprehensive framework for organizing, querying, and managing personal data across multiple sources.
CAP is a lightweight Model Context Protocol (MCP) server that acts as a real-time translation layer between a user's data sources (Gmail, Google Calendar, Notion, Slack, etc.) and AI agents. Instead of dealing with dozens of different APIs and data formats, agents can use a single, consistent interface.
This skill transforms a general-purpose AI agent into a specialized personal data assistant by providing:
- Canonical Schema: Consistent data structures across all sources
- Query Patterns: Pre-built query examples for common use cases
- Security Guidelines: Best practices for handling sensitive data
- Utility Scripts: Production-ready tools for data validation, export, and query building
- 30+ Use Cases: Real-world scenarios demonstrating CAP's capabilities
- SKILL.md: Main entry point with core concepts and usage instructions
- Schema Reference: Complete canonical schema for all 5 shelves (Identity, Comms, Calendar, Docs, Tasks)
- Query Examples: 50+ query patterns with filters and parameters
- Security Guide: Data sensitivity tiers, permissions, and safe handling practices
- Use Cases: 30 common scenarios from daily briefings to project management
generate_briefing.py: Format CAP data into readable daily briefingsvalidate_cap_data.py: Validate data against canonical schema requirementsexport_cap_data.py: Export data to CSV, JSON, or Markdown formatsbuild_query.py: Generate CAP queries from natural language
- Sensitivity tier classification (S1, S2, S3)
- Permission scoping and time-bounded grants
- Human-in-the-loop gates for sensitive operations
- Audit logging and data provenance tracking
- Download the
.skillfile from the releases page - In Manus, go to Settings → Skills
- Click "Add Skill" and select the downloaded file
- The skill will be automatically available for use
- Clone this repository or download the skill directory or copy and paste this repo link to your Openclaw ai agent and tell them to add the skill
- Place the
claw-agent-protocoldirectory in your agent's skills folder - Ensure your agent can read the
SKILL.mdfile and execute Python scripts
Once you've installed the skill and connected your CAP server, you can start using these prompts with your OpenClaw agent. Simply copy and paste these commands—no coding required.
Copy this prompt:
Give me my daily briefing using CAP
What happens: Your agent will fetch today's calendar events, due tasks, and recent communications, then format them into a readable summary.
Copy this prompt:
Show me all high-priority tasks due this week from my CAP data
What happens: Your agent will query your tasks shelf and return all pending high-priority items with due dates in the next 7 days.
Copy this prompt:
Show me unread emails from john@example.com using CAP
What happens: Your agent will query your communications shelf and return all unread emails from that sender.
Copy this prompt:
What's on my calendar this week? Use CAP to get the events
What happens: Your agent will fetch all confirmed calendar events for the next 7 days.
Copy this prompt:
Find all my notes tagged with "project-alpha" using CAP
What happens: Your agent will search your docs shelf for notes with that tag.
Copy this prompt:
Show me all communications with Acme Corp from the last 30 days using CAP
What happens: Your agent will query both your identity shelf (to find Acme Corp contacts) and comms shelf (to get recent messages/emails).
Copy this prompt:
What tasks are currently blocked? Check CAP
What happens: Your agent will query your tasks shelf for items with status="blocked" and show you what's stuck.
Copy this prompt:
Show me all meetings with sarah@example.com this month from CAP
What happens: Your agent will query your calendar shelf filtered by attendee email.
Copy this prompt:
Show me all contacts tagged as VIP in CAP
What happens: Your agent will query your identity shelf for people/orgs with the "vip" tag.
Copy this prompt:
Search my CAP documents for anything related to "quarterly budget"
What happens: Your agent will use the knowledge_search tool to find all docs, notes, and files mentioning that topic.
Copy this prompt:
Help me plan next week. Show me my calendar, pending tasks, and any upcoming deadlines from CAP
What happens: Your agent will aggregate data from multiple shelves (calendar, tasks) to give you a comprehensive weekly overview.
Copy this prompt:
Export all my pending tasks from CAP to a markdown file
What happens: Your agent will query your tasks shelf and use the export script to create a formatted markdown file you can download.
- Be specific about time ranges: "this week", "next month", "last 30 days"
- Mention CAP explicitly: This helps your agent know to use this skill
- Combine shelves: Ask for cross-referenced data like "meetings and related emails"
- Use tags: If you tag your data, you can filter by tags in your prompts
- Ask for exports: Your agent can export data to CSV, JSON, or Markdown formats
claw-agent-protocol/
├── SKILL.md # Main skill instructions (104 lines)
├── LICENSE # MIT License
├── README.md # This file
├── references/
│ ├── schema.md # Canonical schema definitions (118 lines)
│ ├── query_examples.md # Query patterns and filters (232 lines)
│ ├── security.md # Security best practices (297 lines)
│ └── use_cases.md # 30 common scenarios (466 lines)
└── scripts/
├── generate_briefing.py # Daily briefing formatter (48 lines)
├── validate_cap_data.py # Schema validator (299 lines)
├── export_cap_data.py # Data exporter (306 lines)
└── build_query.py # Natural language query builder (255 lines)
Total: 2,146 lines of production-grade code and documentation
CAP organizes personal data into 5 canonical shelves:
| Shelf | URI | Description |
|---|---|---|
| Identity | cap://identity |
People, organizations, contacts |
| Comms | cap://comms |
Messages, emails, threads |
| Calendar | cap://calendar |
Events, availability, meetings |
| Docs | cap://docs |
Notes, files, snippets, bookmarks |
| Tasks | cap://tasks |
Tasks, projects, milestones |
Each shelf has a consistent schema with common metadata (id, timestamps, source, confidence, sensitivity).
This skill supports 30+ common use cases, including:
- Daily briefing generation
- Weekly planning assistance
- Task prioritization
- Meeting preparation
- Email triage
- Client communication history
- Follow-up reminders
- Relationship strength analysis
- Meeting scheduling optimization
- Deliverable tracking
- Cross-reference search
- Meeting notes consolidation
- Document version tracking
- Research thread reconstruction
- Knowledge gap identification
- Time audits
- Focus time protection
- Availability sharing
- Overcommitment detection
- Travel planning integration
- Project status dashboards
- Dependency mapping
- Team workload balancing
- Milestone tracking
- Standup report generation
- Family calendar coordination
- Health appointment tracking
- Financial deadline management
- Personal goal tracking
- Digital life audit
This skill implements security best practices:
- Data Sensitivity Tiers: S1 (public), S2 (internal), S3 (sensitive)
- Permission Scoping: Read/write permissions per shelf
- Time-Bounded Grants: Temporary access for sensitive operations
- Human-in-the-Loop: Confirmation required for destructive actions
- Audit Logging: Append-only logs for critical operations
- Data Provenance: Full traceability to source systems
- Python 3.10+
- Access to a CAP MCP server
- MCP-compatible AI agent (OpenClaw, Claude Desktop, etc.)
Contributions are welcome! This skill is designed to be extended and improved by the community.
- Fork the repository
- Create a feature branch (
git checkout -b feature/amazing-feature) - Commit your changes (
git commit -m 'Add amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request
- Additional utility scripts
- More use case examples
- Connector implementations for new data sources
- Improved query patterns
- Enhanced security features
- Documentation improvements
This skill is licensed under the MIT License. See LICENSE for details.
Jason Fleagle - Chief AI Officer
CAP is created by Jason Fleagle. Jason is a Chief AI Officer and Growth Consultant working with global brands to help with their successful AI adoption and management. He is also a writer, entrepreneur, and consultant specializing in tech, marketing, and growth. He helps humanize data—so every growth decision an organization makes is rooted in clarity and confidence. Jason has helped lead the development and delivery of over 500 AI projects & tools, and frequently conducts training workshops to help companies understand and adopt AI. With a strong background in digital marketing, content strategy, and technology, he combines technical expertise with business acumen to create scalable solutions. He is also a content creator, producing videos, workshops, and thought leadership on AI, entrepreneurship, and growth. He continues to explore ways to leverage AI for good and improve human-to-human connections while balancing family, business, and creative pursuits.
- Inspired by the Model Context Protocol (MCP) standard
Built with ❤️ by a senior software engineer & AI Architect for the AI agent community.