Learn how to use the PRDwise agentic framework
The prdwise repository contains all templates, setup scripts, and agent instructions you need to get started. Repository (Git): https://codeberg.org/PRDwise/prdwise Repository (Mercurial): https://hg.sr.ht/~francisco_reyes/prdwise Choose Your Setup Guide New to prdwise or agentic development? Follow the beginner setup guide Includes Mercurial installation, complete walkthrough, and examples Experienced user? See the quick setup guide Get started in minutes with essential commands and references Adding to an existing project? Learn how to onboard your project Covers integration, VCS setup, and migration strategies What’s new Recent additions to the framework, each with its own guide:
Learning mode shifts the agent from doing your work to teaching you how to do it. You stay in the driver’s seat; the agent asks questions, suggests directions, or reviews what you wrote. It is opt-in and intended only for new projects. What it is A configuration layer on top of the standard prdwise workflow. When enabled, the agent picks a coaching style based on your profile (technical or writing) and adjusts how it responds — questions instead of answers, hints instead of solutions, reviews instead of rewrites.
Three workflows sit between “task created” and “agent starts coding”: architect review, research, and the prep-task wrapper that runs them in order with a gate in between. What they are architect-task — reviews the task spec for soundness: dependencies, risks, ADR candidates, missing context. Stops the line when the spec itself is broken. research-task — gathers patterns, files, symbols, and notes the implementer will need. No code, just findings folded back into the task details. prep-task — runs architect first, then research, with a gate. If architect flags blockers, research does not run until the spec is fixed. When to use them The task is non-trivial and you want to catch spec problems before burning time. The task touches code you do not know well — research surfaces the patterns to follow. You want a single command that does both gated end-to-end: use prep-task. When to skip them The task is small and obvious. You already know the patterns and the spec is tight. You are mid-implementation — these run before code, not during. How to invoke Trigger phrases (per the project’s CLAUDE.md):
T is a small bash helper for moving around the task system: list active tasks, open details, edit, show state, archive, work the backlog. It is a convenience wrapper — the canonical workflow lives in agent-task-rules.md. What it is A shell script at scripts/T that wraps common task operations against tasks/index.md and tasks/details/<id>.md. Run T with no arguments or T h for the current command list; the script’s own help is the source of truth for available commands.
An ADR captures one architecture decision: the context, the choice, the alternatives, and the consequences. prdwise treats ADRs as opt-in — you turn them on per project, and you can backfill them retroactively for existing work. What it is A short markdown file per decision, stored in your project. The create-adr workflow walks the agent through drafting one. The retroactive variant lets you write ADRs after the fact for decisions that are already in the code.