Index / GitClaw (Gitagent) · updated Oct 24, 2026
GitClaw (Gitagent)
open-gitagent/gitclaw · healthy · rank 35 of 50 by stars
A git-native agent framework where identity, rules, memory, tools, and skills are all version-controlled files — fork an agent, branch a personality, git log its memory. Recently hardened with security fixes and a Rust rewrite ('ira') powering a Tauri desktop app.
Facts
Repository
Runtime
Posture
Community
Security breakdown
Composite 78 / 100 · how these are scored
higher is safer
higher is safer
higher is safer
higher is safer
higher is riskier
Model access
Read from the repository, not written by a model · 57 files examined
3 providers · runs locally · custom endpoint
Pinned models
gpt-4o-mini released 2024-07-18 — 2 years old (from the public model catalogue)
Pin last edited 2 months ago
Evidence
Decision
Why choose GitClaw (Gitagent) over OpenClaw?
Why choose this
- Git-native agent definition: identity, rules, and memory are diffable, forkable files
- Recent security hardening (RCE fix, race fixes, leak-free cleanup) with regression tests
- Rust port with Tauri desktop app and Ollama local-model management
Tradeoffs
- Far smaller community (712 stars vs OpenClaw's massive ecosystem)
- No multi-channel messaging (WhatsApp/Telegram/Slack) breadth like OpenClaw
- No Reddit presence and thin third-party sentiment data
Best fit
- Developers who want agent config, memory, and skills versioned in git
- Teams forking/branching agent personalities like code
- Users wanting a CLI + SDK + optional desktop app with local models via Ollama
Avoid if
- You need a large, battle-tested community and plugin registry
- You want zero-setup hosted agents with no API key management
- You need proven sandboxing for untrusted multi-tenant workloads
Evidence is moderate: README and commit history are detailed and show active, security-conscious development, but there is zero Reddit discussion and limited independent web coverage, so community sentiment and adoption claims are uncertain.
AI layer reviewed Oct 5, 2026 · how this is written
Star activity
721 stars today
Overview
GitClaw (branded Gitagent) is a git-native AI agent framework: the agent itself is a git repository, with agent.yaml for config, SOUL.md for personality, RULES.md for constraints, and version-controlled memory/, tools/, skills/, and hooks/ directories. This 'agents as repos' model lets you fork agents, branch personalities, and git log an agent's memory — a fundamentally different abstraction from OpenClaw's channel-centric personal assistant model.
Architecturally it ships as a Node/TypeScript CLI + SDK (with a slim core package split from the voice UI to satisfy supply-chain scanners), plus a newer Rust rewrite called 'ira' that powers a Tauri desktop app (Lyzr Edgespace) with an embedded web UI, Ollama-based local model management, and an onboarding wizard. It supports MCP servers (stdio + http/sse), declarative YAML tools, lifecycle hooks, and an OpenAI-compatible API surface.
Recent commits show serious engineering hygiene: two hardening batches closed a load-time RCE via shell-interpolated git clone, fixed tool concurrency races, wired real abort/steer semantics, and eliminated e2b VM and credential leaks — all with regression tests. Telemetry work groups model requests into per-turn traces. Compared to OpenClaw, GitClaw is more developer-infrastructure than end-user assistant: stronger on auditability and git workflows, weaker on messaging-channel breadth and community scale.