pratikrishi97/councilforge — explained in plain English
Analysis updated 2026-05-18
Turn a GitHub repository or pasted notes into a decision ready requirements document
See a simulated debate between product, engineering, security, and other expert roles before finalizing a plan
Convert a set of requirements into a traceable backlog of epics, stories, and tasks
Generate a phased roadmap with cost and sprint estimates from raw project material
| pratikrishi97/councilforge | 0xradioac7iv/tempfs | 52191314/web-agent-proxy-sdk | |
|---|---|---|---|
| Stars | 0 | 0 | 0 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | moderate | moderate | moderate |
| Complexity | 3/5 | 3/5 | 4/5 |
| Audience | pm founder | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Runs locally as a Next.js app with Bun, can use free AI providers, your own API keys, or a local model.
CouncilForge is a tool that turns raw material, such as a GitHub repository, a product brief, or pasted notes, into finished planning documents like requirement specs, business documents, statements of work, roadmaps, and task backlogs. Its central idea is that instead of asking a single AI prompt to write a document, it runs a simulated debate between several AI roles that each represent a different perspective, such as product, business, architecture, engineering, quality assurance, security, user experience, delivery, and compliance. These roles argue over two rounds, and a moderator role then summarizes where they agree and honestly reports where they still disagree, rather than papering over disagreement. Every important decision made during this process gets logged along with the reasoning behind it, the alternatives that were considered, who supported or objected to it, and how confident the system was. This is meant to let someone later ask why a particular choice was made and get a clear answer, similar to the kind of cross functional review a document would normally get from a real team. Under the hood, it is built as a Next.js web application with a pipeline that runs the whole process in 29 stages, along with a knowledge graph that connects goals, features, requirements, tests, and releases to each other so decisions stay traceable. It can also generate acceptance tests, convert plans into a backlog of stories and tasks, build a phased roadmap, and estimate costs and how many work sprints something will take. The tool supports multiple AI providers rather than locking you into one, with four different modes for connecting to them, including options that use free providers, your own paid API keys, or a locally run model. There is also a separate, code free version of the same council workflow, described as a skill package, that can run inside AI coding tools like Claude Code, Cursor, or GitHub Copilot without needing this project's full application. This project is intended for product managers, business analysts, architects, engineering leads, founders, and people who already use AI coding tools and want a more thorough, multi perspective way of turning messy notes into a defensible plan. It is released under the MIT license.
A tool that turns repos, notes, or briefs into finished planning documents by simulating a debate between multiple AI expert roles, then logging every decision made.
Mainly TypeScript. The stack also includes TypeScript, Next.js, Prisma.
Use, modify, and distribute this code freely, including commercially, as long as you keep the license notice.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly pm founder.
This repo across BitVibe Labs
Verify against the repo before relying on details.