| name | ap-init |
| description | Initialize or re-ground a project by creating durable truth files (PROJECT.md, DECISIONS.md, USERS.md, SECURITY.md, PROGRESS.md, BACKLOG.md, LOADOUT.md, STATUS.md). Use when starting new work or when scope has drifted. |
/ap:init
Purpose
Initialize (or re-ground) a project using agent-pack.
This command creates the durable truth set and the initial working surface:
- PROJECT.md (what we're building)
- DECISIONS.md (durable decisions)
- USERS.md (durable user segments, JTBD, goals, and success signals)
- SECURITY.md (security posture, constraints, and baseline controls)
- PROGRESS.md (completed work and learnings)
- BACKLOG.md (problem list)
- LOADOUT.md (installed modules/loadouts and phase usage guidance)
- STATUS.md (what's happening next)
Use /ap:init when starting a new repo or when scope has drifted.
Inputs
Start a conversation with the user to gather as much as you can. Rough notes are fine, but we can't produce final content without clarity.
- Project idea / description
- Target users
- Constraints (time, budget, tech, compliance)
- Preferred stack (optional)
- Non-goals (what we are not doing)
If inputs are missing, this command must surface them as Unknowns / needs input. If the user cannot provide them, make assumptions but mark them clearly. Ask follow-up questions to clarify vague or incomplete inputs, and iterate until you have enough to proceed. Check with the user before finalizing the output.
Reads
Must consult the following if they exist:
- .agent-pack/core/context/PROJECT.md
- .agent-pack/core/context/DECISIONS.md
- .agent-pack/core/context/USERS.md
- .agent-pack/core/context/SECURITY.md
- .agent-pack/core/context/PROGRESS.md