- 22 AI providers (OpenRouter, Anthropic, OpenAI, Mistral, etc.) - 7 channels (CLI, Telegram, Discord, Slack, iMessage, Matrix, Webhook) - 5-step onboarding wizard with Project Context personalization - OpenClaw-aligned system prompt (SOUL.md, IDENTITY.md, USER.md, AGENTS.md, etc.) - SQLite memory backend with auto-save - Skills system with on-demand loading - Security: autonomy levels, command allowlists, cost limits - 532 tests passing, 0 clippy warnings
1.8 KiB
1.8 KiB
Changelog
All notable changes to ZeroClaw will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
0.1.0 - 2025-02-13
Added
- Core Architecture: Trait-based pluggable system for Provider, Channel, Observer, RuntimeAdapter, Tool
- Provider: OpenRouter implementation (access Claude, GPT-4, Llama, Gemini via single API)
- Channels: CLI channel with interactive and single-message modes
- Observability: NoopObserver (zero overhead), LogObserver (tracing), MultiObserver (fan-out)
- Security: Workspace sandboxing, command allowlisting, path traversal blocking, autonomy levels (ReadOnly/Supervised/Full), rate limiting
- Tools: Shell (sandboxed), FileRead (path-checked), FileWrite (path-checked)
- Memory (Brain): SQLite persistent backend (searchable, survives restarts), Markdown backend (plain files, human-readable)
- Heartbeat Engine: Periodic task execution from HEARTBEAT.md
- Runtime: Native adapter for Mac/Linux/Raspberry Pi
- Config: TOML-based configuration with sensible defaults
- Onboarding: Interactive CLI wizard with workspace scaffolding
- CLI Commands: agent, gateway, status, cron, channel, tools, onboard
- CI/CD: GitHub Actions with cross-platform builds (Linux, macOS Intel/ARM, Windows)
- Tests: 159 inline tests covering all modules and edge cases
- Binary: 3.1MB optimized release build (includes bundled SQLite)
Security
- Path traversal attack prevention
- Command injection blocking
- Workspace escape prevention
- Forbidden system path protection (
/etc,/root,~/.ssh)