This document defines the vision, goals, and constraints for this repository in a way that both humans and Codex agents (like the PM agent) can understand and follow.
The PM agent (pm.md / $pm-round) should always consult this file first when:
- Triaging ADMIN issues
- Assigning priority labels (high / medium / low)
- Proposing new issues with the PM label
Fill in the sections below. Agents will treat them as the single source of truth unless you update this file.
Human (you) should fill this in.
-
One-sentence vision:
Example: "Turn this repo into an autonomous project manager that can triage, plan, implement, and QA work with minimal human oversight." -
Short paragraph (2–4 sentences) describing:
- Who this project is for
- What pain it solves
- What “done” or “success” looks like from your perspective
Define the main outcome you want in the next 1–3 months.
-
Primary goal (1–2 sentences):
Example: "Have a stable set of Codex automations (PM + Dev) that can reliably take an issue from ADMIN → READY → implemented PR with minimal manual intervention." -
Key capabilities required for this goal (3–7 bullets):
- …
- …
These are rules and preferences that shape how work should be done.
-
Quality bar:
Example: "Prefer small, well-scoped changes with clear tests over large, risky refactors." -
Risk tolerance:
Example: "Avoid destructive operations (force pushes, mass file deletions, schema drops) unless explicitly requested by the human." -
Style / UX preferences (if applicable):
Example: "Favor clear, simple UX over cleverness; consistent naming and file structure over ad-hoc organization." -
Communication tone for comments/issues:
Example: "Be concise, neutral, and factual. Avoid over-explaining unless specifically asked."
Clarify what is in scope and out of scope for this repo and its automations.
-
In scope (what we want more of):
- …
- …
-
Out of scope (what we should avoid or defer):
- …
- …
-
high:
- Directly unblocks the North Star goal or a core capability
- Fixes a critical bug or broken workflow
- Prevents data loss, security problems, or major user pain
-
medium:
- Clearly improves the system in a noticeable way
- Adds or refines capabilities that support the main workflows
- Addresses non-critical bugs or UX rough edges
-- low:
- Nice-to-have improvements or polish
- Experiments, ideas, or refactors that are not currently blocking anything
Group future work into a few themes so agents can cluster proposals.
For each theme you care about: Name: Short description (1–2 sentences): Example tasks:
List any hard constraints or important external systems.
Tooling constraints (languages, runtimes, services that must or must not be used) Integration points (e.g. GitHub, specific APIs, third-party services) Performance / cost constraints (e.g. "avoid running very expensive CI jobs repeatedly")