This page documents the history and development timeline of MicroClaw, a Rust-based agentic AI assistant.
MicroClaw is an agentic AI assistant built with Rust that lives in chat surfaces. The project was:
| Metric |
Value |
| Total Commits |
689+ |
| Current Branch |
main |
| Contributors |
9+ |
| GitHub Stars |
459+ |
| Forks |
71+ |
| Watchers |
13+ |
MicroClaw has published 93+ releases with rapid iteration:
| Version |
Release Date |
Notes |
| v0.0.128 |
February 28, 2026 |
Latest release |
| v0.0.127 |
February 2026 |
IRC channel support |
| v0.0.120+ |
January-February 2026 |
Multiple stability improvements |
| … |
2025-2026 |
90+ earlier releases |
The project follows a rapid release cycle with frequent updates and improvements.
- Agentic AI Loop - Full tool execution with up to 100 iterations per request
- Multi-Platform Support - Telegram, Discord, Slack, Feishu/Lark, Matrix, WhatsApp, iMessage, Email, Nostr, Signal, DingTalk, QQ, IRC, and Web
- Multiple LLM Providers - Anthropic (native) + OpenAI-compatible APIs
- Persistent Memory - AGENTS.md files + SQLite structured memory
- Tool System - 20+ built-in tools
- MCP Support - Model Context Protocol servers
- Plugin System - Manifest-based plugins
- Plan & Execute - Todo list tools for multi-step tasks
- Scheduled Tasks - Cron-based recurring tasks
- Local Web UI - Cross-channel history viewer
Rust Crate Structure:
crates/
├── microclaw-core/ # Shared error/types/text modules
├── microclaw-storage/ # SQLite DB + memory domain + usage reporting
├── microclaw-tools/ # Tool runtime primitives + sandbox + helper engines
├── microclaw-channels/ # Channel abstractions and routing boundary
└── microclaw-app/ # App-level support modules
Main Application Structure:
src/
├── main.rs # CLI entry point
├── runtime.rs # Runtime bootstrap + adapter startup
├── agent_engine.rs # Channel-agnostic agent loop
├── llm.rs # Provider abstraction
├── channels/*.rs # Concrete channel adapters
├── tools/*.rs # Built-in tool implementations
├── scheduler.rs # Background scheduler and reflector loop
└── web.rs # Web API + stream endpoints
- IRC Support - Added IRC channel adapter with TLS and SASL support
- Feishu/Lark - Chinese collaboration platform support
- Matrix - Decentralized chat protocol support
- Multi-Account - Support for multiple accounts per platform
-
Apple Ecosystem Skills:
apple-notes - Apple Notes integrationapple-reminders - Apple Reminders integrationapple-calendar - Apple Calendar integration
-
Browser Automation:
- Playwright MCP with extension mode
- Chrome 136+ support
-
Desktop Automation:
- macOS: Peekaboo MCP integration
- Windows: pywinauto MCP support
- Interactive Setup Wizard -
microclaw setup command
- Gateway Service Mode - Persistent deployment as system service
- Diagnostics -
microclaw doctor command with JSON output
- One-Line Installer - Automated installation script
- Docker Sandbox Mode - Three security profiles (hardened, standard, privileged)
- Path Isolation - Shared vs. chat-scoped working directories
- High-Risk Tool Confirmation - Required confirmation for dangerous operations
- Multi-Chat Permission Model - Control chat restrictions
- Semantic Retrieval - Optional sqlite-vec embeddings
- Quality Gates - Memory quality filtering
- Lifecycle Management - Soft-archive and confidence fields
- Memory Observability - Reflector throughput and injection coverage metrics
MicroClaw uses a channel-agnostic core with platform adapters, enabling:
- One shared agent loop for all platforms
- Consistent tool system across channels
- Unified storage and memory layer
- Easy addition of new platform adapters
Full conversation state persistence in SQLite:
- Tool interactions preserved
- Context compaction via summarization
- Automatic session state management
¶ Rate Limit Handling
- Exponential backoff on 429 rate limits
- 3 retry attempts by default
- Provider-specific rate limit handling
- Channel-specific message splitting
- Telegram: 4096 character limit
- Discord: 2000 character limit
- Slack: 4000 character limit
- Continuous typing indicator during processing
¶ Documentation Standards
MicroClaw maintains anti-drift documentation:
| Document |
Description |
docs/generated/tools.md |
Auto-generated tool reference |
docs/generated/config-defaults.md |
Configuration defaults |
docs/generated/provider-matrix.md |
LLM provider compatibility matrix |
| Document |
Description |
README.md |
Overview, setup, usage |
DEVELOP.md |
Developer guide - architecture, adding tools |
TEST.md |
Manual testing guide |
CLAUDE.md |
Project context for AI coding assistants |
AGENTS.md |
Agent-friendly project reference |
¶ Published Articles
- “Building MicroClaw: An Agentic AI Assistant in Rust That Lives in Your Chats” - Technical deep-dive into architecture and design decisions
The official blog at microclaw.ai/blog covers:
- Release announcements
- Feature deep-dives
- Integration guides
- Best practices
¶ Standards & Protocols
MicroClaw supports the Anthropic Skills standard:
- Auto-discovered skills from
<data_dir>/skills/
- Compatible with Anthropic skill format
- Extensible custom skill creation
- Protocol Version: 2025-11-05 (default)
- Transports: stdio, streamable_http
- Server Types: Local (stdio), Remote (HTTP)
Status: ✅ Under Active Development
The project is actively maintained with:
- Frequent releases (93+ to date)
- Multiple active contributors
- Regular feature additions
- Responsive issue tracking
Based on recent development activity:
- MCP Ecosystem - Expanding integrations with external tool servers
- Desktop Automation - Cross-platform desktop control
- Memory Enhancements - Semantic retrieval and quality improvements
- Platform Expansion - Additional chat platform adapters
- Developer Experience - Setup wizards, diagnostics, documentation
The project has grown to include:
- 9+ core contributors
- Community contributions via PRs
- Active issue tracking and discussions
MicroClaw is used for:
- Personal infrastructure automation
- Team operations bots
- Product prototyping
- Chat-based development assistance
While inspired by nanoclaw, MicroClaw differentiates itself with:
| Feature |
nanoclaw |
MicroClaw |
| Language |
Python |
Rust |
| Deployment |
Script |
Single binary |
| Memory |
Basic |
SQLite + AGENTS.md + semantic |
| Platforms |
Limited |
14+ platforms |
| MCP Support |
No |
Yes |
| Skills System |
Basic |
Anthropic-compatible |
| Sandbox Mode |
No |
Docker-based |
| Web UI |
No |
Yes |
Any questions?
Feel free to contact us. Find all contact information on our contact page.