CFactory — the control tower for your software factory

CFactory is the agentic cockpit that sits above the PARR pipeline. It threads every unit of work across three autonomous services — planning, execution and verification — into one pane of glass, and adds an LLM copilot that explains what is happening and proposes the next move.

Three autonomous tools are powerful. Three autonomous tools you can see, correlate and steer from one place are a factory.

CFactory is the newest member of the Factory family — and the piece that turns the others into a suite.

The problem it solves

The Factory family already plans, builds and verifies software autonomously:

PFactory  →  AIFactory  →  TFactory
 (Plan)        (Act)       (Reflect/Review)

But each runs on its own portal, its own workspace, its own status model. Nobody can answer the obvious question — “where is feature X right now, and why is it stuck?” — without opening three tabs and stitching the story together by hand.

CFactory answers it. It is the observability + control layer the modern agentic SDLC needs.

What CFactory does

• Threads work across services. A single WorkItem, keyed by GitHub issue, follows the chain plan → code → branch/PR → tests so you always know where something is.
• One live cockpit. A read-first board shows every WorkItem across the plan / code / test stages with status, gates, verdicts and PR links — fed by REST, WebSocket and webhooks from the three services.
• An agentic copilot. Ask “why is #182 stuck?” and get an answer grounded in real cross-service state. The copilot summarises timelines, computes cost and latency rollups, and flags anomalies (stuck phases, runaway handback loops, gate failures, cost spikes).
• Advise + confirm, never silent. The copilot can prepare actions — approve a gate, trigger a handoff, kick a handback — but every write waits for an explicit human click. Human-in-the-loop by design.

Where it fits

See the architecture →   ·   See the roadmap →   ·   Meet the family →

CFactory is early and built in the open. Follow along on GitHub or read the blog.