Red Queen brings deterministic pipeline orchestration to AI coding agents
Red Queen is an open-source, YAML-configured orchestration layer that shuttles tickets through a deterministic state machine pipeline — spec → plan review → code → review → test → human review — without spending any AI tokens on routing.
Score breakdown
Red Queen addresses a gap the source identifies — the lack of a deterministic, auditable pipeline layer above existing AI coding agents — by providing token-free routing, configurable human gates, and retry-with-escalation logic as first-class workflow primitives.
- 01Red Queen is a YAML-configured, deterministic state machine orchestrator for AI coding agent pipelines.
- 02The orchestrator spends zero AI tokens on routing — all token spend goes to actual code work.
- 03Default pipeline phases: spec writing → plan review → coding → code review → testing → human review.
Red Queen positions itself as "the Jenkins for your AI coding workflow," describing its relationship to Claude Code the way Jenkins relates to bash. The core idea is a deterministic state machine that routes tickets through a configurable sequence of phases — spec writing, plan review, spec feedback, coding, code review, testing, and human review — without any LLM involvement in the routing layer itself. All orchestration logic is pure deterministic code, meaning every AI token spent goes to actual work rather than coordination decisions.
The pipeline is defined in `redqueen.yaml`, where users can add, remove, or reorder phases and human-in-the-loop gates.
The pipeline is defined in `redqueen.yaml`, where users can add, remove, or reorder phases and human-in-the-loop gates. Each phase runs a purpose-built, isolated prompt worker (spec writer, coder, reviewer, tester, comment-handler), and skills are markdown prompt templates that follow the agentskills.io format. Users can override any built-in skill by placing a `SKILL.md` file in `.redqueen/skills/`. Failed phases retry up to three times before escalating to a human gate, guarding against runaway token usage. Optional features include a webhook server for instant response and a polling reconciler that requires no external queue.
Red Queen explicitly distinguishes itself from adjacent tools: it is not an agent host (it spawns `claude` CLI subprocesses rather than replacing the agent runtime), not a skill registry, not a spec framework, and not an interactive coding tool like Claude Code or Aider. It is installed via `npm install -g redqueen` and integrates bidirectionally with GitHub Issues and Jira, with Linear listed as on the roadmap. An adapter pattern means new issue trackers or source control systems require only a new adapter implementing the `IssueTracker` or `SourceControl` interface.
Key facts
- 01Red Queen is a YAML-configured, deterministic state machine orchestrator for AI coding agent pipelines.
- 02The orchestrator spends zero AI tokens on routing — all token spend goes to actual code work.
- 03Default pipeline phases: spec writing → plan review → coding → code review → testing → human review.
- 04Failed phases retry up to three times, then escalate to a human gate to prevent runaway token bills.
- 05Skills are markdown prompt templates; users override built-ins by dropping a `SKILL.md` into `.redqueen/skills/`.
- 06Bidirectional sync with GitHub Issues and Jira is supported; Linear is listed as on the roadmap.
- 07Installed via `npm install -g redqueen`; dispatches `claude` CLI subprocesses as workers.
Topics
Summary and scoring are generated automatically from the original article. We always link back to the publisher and never republish images or paywalled content. Last processed Jun 10, 2026 · 15:34 UTC. How this works →