nanoclaw-repl
affaan-m/ecc
Session-aware REPL for Claude with persistent markdown storage and dynamic skill loading.
What is nanoclaw-repl?
NanoClaw v2 is a zero-dependency REPL built on claude -p that manages persistent markdown-backed sessions. Use it to maintain long-running conversations, switch models, load skills dynamically, and branch sessions for experimentation.
- Persist sessions as markdown files for easy version control and inspection
- Switch between Claude models mid-session with /model
- Load and unload skills dynamically with /load
- Branch sessions to explore alternatives without losing history
- Search across multiple sessions with /search
- Compact session history to reduce file size with /compact
How to install nanoclaw-repl
npx skills add null --skill nanoclaw-repl- Node.js installed
- Access to scripts/claw.js in the ECC repository
- Claude API credentials configured for claude -p
How to use nanoclaw-repl
- 1.Run scripts/claw.js to start the REPL
- 2.Use /model <model-name> to switch Claude models
- 3.Use /load <skill-name> to dynamically load skills
- 4.Use /branch <name> to create a branching point before risky changes
- 5.Use /search <query> to find content across sessions
- 6.Use /compact to reduce session file size after milestones
- 7.Use /export <format> to save sessions as md, json, or txt
- 8.Use /metrics to view session statistics
Use cases
- Maintaining long-running development sessions that survive restarts
- Experimenting with different Claude models on the same conversation
- Branching a session to test risky changes before committing
- Searching conversation history across multiple related sessions
- Archiving completed work by exporting to shareable formats
- Developers extending or operating ECC's NanoClaw REPL
- Teams managing persistent Claude conversations
- Researchers tracking multi-model experiments
- Anyone needing version-controlled conversation history
nanoclaw-repl FAQ
Sessions are stored as markdown files, making them human-readable, version-controllable, and compatible with standard tools.
Yes, use /model <model-name> to switch between Claude models while preserving conversation history.
Branching creates a checkpoint of your current session so you can explore alternatives without losing the original conversation path.
No, NanoClaw v2 is zero-dependency by design, requiring only Node.js and Claude API access.
Use /export <format> to convert your session to markdown, JSON, or text, then share or store the exported file.
Full instructions (SKILL.md)
Source of truth, from affaan-m/ecc.
name: nanoclaw-repl description: Operate and extend NanoClaw v2, ECC's zero-dependency session-aware REPL built on claude -p. metadata: origin: ECC
NanoClaw REPL
Use this skill when running or extending scripts/claw.js.
Capabilities
- persistent markdown-backed sessions
- model switching with
/model - dynamic skill loading with
/load - session branching with
/branch - cross-session search with
/search - history compaction with
/compact - export to md/json/txt with
/export - session metrics with
/metrics
Operating Guidance
- Keep sessions task-focused.
- Branch before high-risk changes.
- Compact after major milestones.
- Export before sharing or archival.
Extension Rules
- keep zero external runtime dependencies
- preserve markdown-as-database compatibility
- keep command handlers deterministic and local
Related skills
More from affaan-m/ecc and the wider catalog.
nasiko-control-plane
Agent skill from affaan-m/ecc.
nestjs-patterns
Production-grade NestJS patterns for modular TypeScript backends with validation, guards, and config.
netmiko-ssh-automation
Safe Python Netmiko patterns for read-only SSH collection, bounded batch automation, and guarded config changes.
network-bgp-diagnostics
Read-only BGP troubleshooting: diagnose neighbor state, route exchange, and policy issues safely.
network-config-validation
Pre-deployment validation for Cisco IOS/IOS-XE configs: dangerous commands, IP conflicts, stale references, and security hygiene.
network-interface-health
Diagnose physical link, duplex, and congestion issues by analyzing interface error and drop counters on routers, switches, and Linux hosts.