Index / MicroClaw · updated Aug 17, 2026
MicroClaw
microclaw/microclaw · healthy · rank 27 of 43 by stars
A Rust-native, channel-agnostic agent runtime focused on rock-solid reliability and verified task completion. Aggressive active development with strong emphasis on sandboxing, governance, and operational resilience.
Facts
Repository
Runtime
Posture
Security breakdown
Composite 88 / 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 · 33 files examined
2 providers · custom endpoint
Pinned models
claude-sonnet-4-5-20250929 released 2025-09-29 — 11 months old (dated in the model id)
Pin last edited 4 days ago
Evidence
Decision
Why choose MicroClaw over OpenClaw?
Why choose this
- Rust-native performance and low memory footprint (micro-optimized)
- Built-in verified completion contracts with evidence stamping
- Strong governance: tool policy gates, token budgets, editable config
Tradeoffs
- Smaller community and plugin ecosystem vs OpenClaw
- Main branch unstable; stable branch lags
- Less cross-platform mobile support (no native Android/iOS apps mentioned)
Best fit
- Self-hosters wanting a Rust-based multi-channel agent
- Teams needing verified task completion and governance
- Users wanting extensible skills/MCP with local control plane
Avoid if
- Those needing a stable production release (main is volatile)
- Users wanting extensive community plugins (ecosystem emerging)
- Non-technical users uncomfortable with CLI/config
Evidence from README and recent commits is strong for architecture and security, but community sentiment is uncertain due to zero Reddit mentions and weak web footprint specific to MicroClaw.
AI layer reviewed Aug 2, 2026 · how this is written
Star activity
731 stars today
Overview
MicroClaw is a Rust-built, channel-agnostic agent runtime that unifies Telegram, Discord, Slack, Feishu, IRC, and web interfaces under a single persistent agent loop. It emphasizes agentic execution with multi-step tool use, sub-agents, scheduled tasks, skills, and MCP support, while providing a local web control plane for governance and task management.
Recent development focuses heavily on reliability and security: completion contracts verify task outcomes with machine-checkable evidence, a tool-policy gate enforces pre-call restrictions, ClawHub installs are tree-pinned and injection-scanned, and provider circuit breakers with fallback models improve resilience. The architecture includes delivery outboxes for guaranteed message delivery, interrupted-turn recovery, and supervised background loops with restart backoff.
Compared to OpenClaw, MicroClaw prioritizes a lightweight Rust footprint (micro-optimized memory/boot) and operational rigor, but trades off a smaller community, less mobile-native support, and a rapidly changing main branch that recommends using the stable branch for production.