Summary

The NanoClaw REPL skill provides instructions for operating and extending a session-aware REPL environment (`scripts/claw.js`) with features like persistent markdown sessions, model switching, dynamic skill loading, and session branching. It emphasizes zero external dependencies and markdown-as-database compatibility as core design principles.

Detected Capabilities

Trigger Keywords

Phrases that agents use to match this skill to user intent.

Risk Signals

Use Cases

• Run interactive REPL sessions with persistent markdown-backed history
• Switch between different Claude models within a session
• Load and manage dynamic skills during REPL operation
• Branch sessions before attempting high-risk changes
• Search and cross-reference content across multiple sessions
• Compact session history after completing major milestones
• Export session data to markdown, JSON, or text formats for sharing or archival
• Monitor and analyze session performance metrics

Quality Notes

• Skill provides clear capabilities list and operating guidance
• Extension rules document design constraints but lack implementation details
• Minimal documentation of error handling or edge cases (e.g., corrupted markdown, missing sessions)
• No examples of command syntax or expected outputs
• Scope is clear at a high level but lacks detail on file paths, session storage location, and session format specification
• No guidance on handling large sessions or performance considerations
• Extension rules are sound (zero dependencies, deterministic handlers) but lack rationale or examples