Sagart-cactus/SprintFoundry — reverse-engineered prompt
Reverse engineered prompt
Build me a TypeScript command line app called SprintFoundry that can take a GitHub issue, Linear ticket, Jira ticket, or a plain prompt, turn it into a plan, then have different AI roles work through it until there is tested code ready as a pull request.
I want it to feel like a small AI software team. There should be roles like product, architect, developer, QA, security, UI and DevOps. The tool should validate the plan, run the right agents in order, keep their work in separate workspaces, save logs and artifacts, and support human approval gates when needed.
Please include a simple setup flow, project config files, commands to run a task, inspect logs, resume a failed run, and run one agent directly. Add a local monitor web page that shows live progress, step results, streaming output, token usage, and cost. Also include a Docker based distributed mode with Postgres and Redis. Look up current docs online if needed.
Want more depth? Deep Reverse