Aionrs

aionrs

A Rust-based LLM tool-use agent for the command line. It connects to LLM APIs, autonomously invokes local tools (file I/O, shell, search, etc.), and completes tasks end-to-end.

Features

• Multi-provider — Anthropic, OpenAI (and compatibles like DeepSeek/Ollama/Gemini), AWS Bedrock, Google Vertex AI
• ProviderCompat layer — Configuration-driven compatibility for provider quirks (no hardcoded conditionals)
• Reasoning model support — Configurable thinking and reasoning effort for supported models
• Local tool suite — Read, Write, Edit, ExecCommand, Grep, Glob, ViewImage, and Spawn (sub-agents), plus Skills, plan mode, and deferred tool discovery
• OpenAI Responses API — Use either Chat Completions or Responses through the compatibility layer
• MCP client — Connect to any Model Context Protocol server (stdio / SSE / streamable-http)
• Dynamic MCP injection — Host clients can inject MCP servers at runtime via the JSON stream protocol
• Skills — Named prompt snippets with variable substitution, shell expansion, conditional activation, and per-skill model/permission overrides (see docs/skills.md)
• Hook system — Event-driven automation on tool lifecycle (auto-format, lint, audit)
• Sub-agent spawning — Parallel task execution via the Spawn tool
• Session persistence — Save and resume conversation history
• Persistent memory — Project-specific memory with auto-indexing across sessions (see docs/advanced.md)
• Plan mode — Read-only exploration mode for designing implementation plans before coding (see docs/advanced.md)
• Context compression — Stable tool-output limits, token-based autocompact, and an emergency guard (see docs/advanced.md)
• Output compaction — Configurable output compression (off/safe/full) with T