document-existing-project

name	document-existing-project
description	Documents an existing codebase in Meridian phase docs and as-is inventory without creating epics or user stories. Use with /document-project for brownfield adoption or refreshing docs from code.
allowed-tools	Read, Glob, Grep, Bash, Edit, Write

Document existing project (Meridian)

Captures current state in docs/ — inventory + phase docs 00–08. No epics, versions, sprints, or US.

Selective reading

File	When to read
.agent/references/templates/init-interview-guide.md	Mandatory — interview gate (Mode B bank)
.agent/references/templates/as-is-inventory-template.md	Mandatory before as-is.md
.agent/skills/init-project/references/doc-templates.md	Mandatory — flow + depth bar
.agent/references/templates/phase-docs/*.md	Mandatory — one per phase doc you Write
docs/discovery/product-brief.md	If present — align intent
Existing docs/*.md	Merge, do not blindly overwrite approved content

When to trigger

• Existing codebase adopting Meridian (/init-meridian Mode B next step)
• docs/ exists but phase docs are thin or empty
• Manager asks: document as-is, baseline, inventory, brownfield, “what we have today”
• Workflow /document-project

Do not use for: /create-us, retroactive ✅ on legacy work, or marking docs approved.

---

Preconditions

Check	Action
docs/ missing	Run /init-meridian first (structure + bootstrap only)
.agent/ present	Required
Manager authorization	Confirm target package root

---

Procedure

Phase 0 — code reading (before interview)

1. Manifest files: package.json, pyproject.toml, Cargo.toml, go.mod, etc.
2. Top-level structure — apps, packages, src/.
3. README, existing docs, ADRs.
4. Auth, routes, models, DB configs, CI workflows.

Phase 1 — interview

Follow init-interview-guide.md Mode B. Lead with inferences.

Phase 2 — as-is inventory

Create or refresh docs/inventory/as-is.md:

• One row per user-facing capability
• Evidence = paths, routes, tests
• Confidence high | medium | low
• Epic candidates as labels only — do not create EPIC-XX rows

Phase 3 — populate phase docs

For each docs/0X_*.md missing or thinner than depth bar, read matching phase-docs/*.md and Write:

Doc	Source
00_scope	README + inventory + interview; include Current product state
01_tech_stack	Manifests + folder structure
02_security	Auth libs, env, write scopes
03_user_types	Roles, routes, user models
04_principles	Infer layers from repo; mark (review) where uncertain
05_architecture	Actual structure — draft, not approved
06_database	Schema, migrations, SQLite paths
07_api_contracts	Routes, GraphQL, CLI — or _n/a_
08_environments	README setup + CI

Keep status: draft. If a section was approved, only append Assumptions or propose edits in output — do not downgrade status without manager ask.

Phase 4 — cross-check

• Every high inventory row appears in 00_scope or 05_architecture
• 03_user_types matches 00_scope audience
• 04_principles layer table matches 05 layers
• Run validate_meridian.py <root> --sqlite-only

Phase 5 — decisions

update-decisions-log if documentation materially changed product understanding.

---

Prohibitions

Forbidden	Allowed
Create US / epics / versions	Inventory + phase docs
status: approved on any doc	draft + assumptions
Retroactive US with ✅	Optional epic candidates in inventory only
Delete manager-approved content without ask	Merge + gap notes

---

Output

Document project complete:
Package root:
Inventory: created | updated | unchanged
Phase docs updated: [list paths]
High-confidence capabilities: N
Assumptions requiring human review:
Gaps (thin sections):
Next: human review 00→04 | /audit-docs | /security-pass | /architecture