autodev

autonomous software-development loop for oh-my-pi

中文版文档 · Architecture · Testing · Design Decisions · Installation · Changelog

autodev is a meta-agent that orchestrates LLM subagents through a YAML-persisted state machine to complete coding tasks autonomously. It does not directly modify files or run commands -- it operates a state machine that does.

Note

Early stage. Core state machine is stable and tested (Axis 1). LLM prompt adherence is being optimized via adversarial testing (Axis 2). Issues and PRs welcome.

Quick Start

# 1. Install (project-level)
tar -xzf autodev-v{VERSION}-offline.tar.gz -C .omp/

# 2. Restart omp or reload extensions

# 3. Start
/autodev "add user authentication to the API"

For detailed installation options, see INSTALL.md.

Architecture at a Glance

autodev runs a five-phase loop, each phase enforced by the state machine:

flowchart LR
    RECON-PLAN --> RECON --> PLAN --> EXECUTE --> FINAL --> DONE
    RECON -.->|low confidence| RECON-PLAN
    EXECUTE -.->|blocked| EXECUTE

Loading

• RECON-PLAN: LLM decides what to investigate
• RECON: Isolated subagents per dimension return structured dossiers
• PLAN: Two-round adversarial gate review (draft + critique)
• EXECUTE: Sliced topological execution: Design -> Implement -> Verify
• FINAL: Build standard + final check

Full architecture diagram and details >

Key Features

• YAML-persisted state machine -- crash recovery via resume anchor; all state changes auditable
• TwoRoundGate -- adversarial acceptance criteria review (R1 draft, R2 loophole finding)
• Dynamic recon dimensions -- LLM decides what to investigate per task, not hardcoded templates
• HITL/HOTL/auto -- three modes, same core loop, zero divergence
• Context budget guardrails -- hard tool-level gate prevents context overflow
• Slice boundary handoff -- prevents context rot across long tasks

Learn More

Document	What's Inside
Architecture	Full loop diagram, all 8 design features, project structure
Testing	Two-axis methodology, coverage matrix, running tests
Design Decisions	Why YAML? Why separate tools? Why hard gates? Limitations
Installation	Global vs project-level install, verification, uninstall

License

MIT