orchestrate

Multi-agent orchestration automates end-to-end software development by running a nine-stage pipeline from brainstorming to PR-ready artifacts. You provide a prompt or spec file, and the pipeline handles planning, coding, testing, and version control through isolated sub-agents.

Multi-agent pipeline orchestration coordinates isolated sub-agents to handle distinct development stages like Q&A, planning, coding with TDD, and quality gates. It enforces stage isolation, preserves worktrees, and extracts artifacts to ensure rigorous failure handling throughout the workflow.

To start an automated development pipeline, provide a spec.md file path or a TASK_CONTEXT prompt. The orchestrator initiates the nine-stage workflow, automatically propagating task context and generating PR-ready results and tracked artifacts in docs/spec/<name>/.

Yes, automated coding pipelines can run parallel execution where possible and apply Test-Driven Development (TDD) during the coder stage. The orchestration includes stagnation detection and a quality gate to ensure code meets requirements before proceeding.

Multi-agent orchestration is designed for end-to-end software projects requiring brainstorming, Q&A, planning, coding, testing, documentation, and version control. It syncs docs and captures learnings automatically as part of the nine-stage workflow.

Multi-agent orchestration enforces rigorous failure handling by isolating stages and extracting artifacts to docs/spec/<name>/ throughout the workflow. This ensures intermediate outputs like plans and code are preserved even if a subsequent stage fails.