nanoclaw-repl

Operate and extend NanoClaw v2, a zero-dependency session-aware REPL built on claude -p.

1|Updated Oct 11, 2025
One-click install
npx skills add https://github.com/ibytechaos/claude --skill nanoclaw-repl-ibytechaos
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: nanoclaw-repl
Source: https://github.com/ibytechaos/claude/tree/main/plugins/everything-claude-code/skills/nanoclaw-repl
Command: npx skills add https://github.com/ibytechaos/claude --skill nanoclaw-repl-ibytechaos

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Managing long-running AI coding sessions often leads to lost context, unorganized history, and no way to branch or recover work. This Skill provides operational guidance for NanoClaw v2, a session-aware REPL that persists conversations as markdown files. ## Core Features & Use Cases - Persistent Sessions: Markdown-backed session storage with branching via /branch and cross-session search via /search. - Session Lifecycle Management: Switch models with /model, load skills dynamically with /load, compact history with /compact, and export to md/json/txt with /export. - Use Case: Before attempting a risky refactor, branch your current session, then compact history after hitting a milestone and export the session as markdown for archival or sharing with teammates. ## Quick Start Ask the assistant to start a NanoClaw session, branch it before a risky change, and export the result as markdown.

Frequently Asked Questions about nanoclaw-repl

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I manage persistent AI coding sessions with NanoClaw?

NanoClaw stores sessions as markdown files, so history persists across runs. Use /branch to fork a session before risky changes, /compact to shrink history after milestones, and /export to save sessions as md, json, or txt.

How to switch models or load skills in a NanoClaw REPL session?

Use the /model command to switch the underlying model mid-session and /load to dynamically load a skill into the active session. Both commands run locally without restarting the REPL.

Does NanoClaw require external dependencies to run?

No. NanoClaw v2 is built on claude -p with zero external runtime dependencies, and its extension rules require keeping it that way. Sessions are stored as plain markdown files acting as the database.

Can I search across multiple NanoClaw sessions?

Yes. The /search command performs cross-session search over the markdown-backed session store, letting you find prior conversations without manually opening files.

What are the rules for extending NanoClaw with new commands?

Extensions must keep zero external runtime dependencies, preserve markdown-as-database compatibility, and keep command handlers deterministic and local. Session metrics are available via /metrics to monitor behavior.