CrowClaw banner
subinium subinium

CrowClaw

Development community

Description

🐦‍⬛🦞 A self-improving TypeScript agent framework that learns from every conversation

Installation

This entry records only its repository, not the path inside it, so there is no exact command to give. Open the source below and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

README

CrowClaw

CrowClaw

TypeScript runtime for self-improving agents.

npm MIT License Node 22+ 19 packages Tests Changelog


**Beta. Single-maintainer, moving fast.** Pin exact versions. The agent loop and security surface are well-tested (3,575 tests as of v0.9.0, 24-issue v0.9.0 Hermes v0.12/v0.13 parity wave 1 + 17-issue v0.8.4 audit-debt closure + 52-issue v0.8.3 GitHub-close pass + 53-issue v0.8.2 audit + parity sweep + 10-issue v0.8.1 dashboard overhaul + 11-issue v0.8.0 Hermes parity sweep + 18-issue v0.7.1 dashboard audit + 26-issue v0.6.1 follow-up + 103-issue v0.6.0 sweep + 38-issue v0.5.0 sweep). Several subsystems are still partial — see [Feature status](#feature-status).

CrowClaw gives you an agent loop, 50+ tools, skill learning, scheduled jobs, multi-channel webhooks, and a dashboard — without wiring the whole stack yourself.

What it is

CrowClaw is for building **backend agents** that can:

  • run multi-turn tool loops with retries, fallbacks, and approval gates,
  • learn reusable skills from completed conversations and inject them into future runs,
  • expose a CLI, web dashboard, HTTP API, and messaging webhooks,
  • run primarily on Node.js, with a