project-analyzer

project-analyzer

When to use

Use this skill when:

• Starting work on an unfamiliar project
• Onboarding to a new codebase
• Auditing the current state of agent docs, contexts, and features
• Creating a baseline understanding of the project for future work
• Generating comprehensive project documentation for knowledge transfer

Do NOT use when:

• Small, focused code changes
• Regular feature development

Procedure: Analyze a project

A project analysis is a systematic walkthrough of the entire codebase that:

1. Detects — framework, language, tech stack, patterns, legacy vs. modern
2. Inventories — modules, services, models, endpoints, tests
3. Analyzes — business domains, data flows, API contracts, dependencies
4. Documents — writes structured analysis files to agents/evidence/analysis/
5. Assesses — identifies gaps, technical debt, missing docs

It orchestrates other skills and commands to produce a comprehensive picture.

procedure.yaml holds the order: run phases one at a time, each only after the previous met its exit. A judgement exit is sequenced, not verified: state your decision first. Phase 3–6 without its layer writes not-applicable.md (Phase 6: service-map.md) saying why — an absent layer is a finding, not a skip.

1. Phase 1: Project overview — exit: check: analysis-overview-matches-manifests
2. Phase 2: Architecture — exit: exists: agents/evidence/analysis/architecture
3. Phase 3: Data layer — exit: exists: agents/evidence/analysis/models
4. Phase 4: Business domains — exit: exists: agents/evidence/analysis/domains
5. Phase 5: API surface — exit: exists: agents/evidence/analysis/api
6. Phase 6: Service map — exit: exists: agents/evidence/analysis/services/service-map.md
7. Phase 7: Module inventory (if modules exist) — exit: judgement
8. Phase 8: Infrastructure & testing — exit: exists: agents/evidence/analysis/testing/test-map.md
9. Phase 9: Agent docs audit — exit: judgement
10. Phase 10: Gap analysis & action plan — exit: judgement

Analysis output

All analysis results are written to agents/evidence/analysis/ in a structured directory layout. The goal: someone could rebuild the project from these documents alone.

Directory structure

agents/evidence/analysis/
├── overview.md                  ← Project profile, tech stack, architecture summary
├── architecture/
│   ├── database.md              ← Schema, connections, multi-tenancy, migrations
│   ├── api.md                   ← Versioning, routes, middleware, auth flow
│   ├── infrastructure.md        ← Docker, CI/CD, deployment, AWS, monitoring
│   └── patterns.md              ← Design patterns used (Repository, Service, Pipeline, etc.)
├── domains/
│   ├── {domain}.md              ← One file per business domain (see below)
│   └── ...
├── modules/
│   ├── {module}.md              ← One file per module (see below)
│   └── ...
├── models/
│   ├── api-database.md          ← All api_database models, relationships, key columns
│   └── customer-database.md     ← All customer_database models, relationships, key columns
├── services/
│   └── service-map.md           ← All services with purpose, dependencies, key methods
├── api/
│   ├── endpoints-v1.md          ← All v1 endpoints: route, controller, request, resource
│   ├── endpoints-v2.md          ← All v2 endpoints: route, controller, request, resource
│   └── contracts.md             ← API contracts: request/response shapes, validation rules
└── testing/
    └── test-map.md              ← Test suites, coverage areas, test data strategy

Domain analysis files

Each business domain gets its own file in agents/evidence/analysis/domains/. A domain groups related models, services, controllers, jobs, and events around a business concept:

Domain	What it covers
projects.md	Construction sites, positions, project status, geocoding
planning.md	Appointments, crew assignments, capacity planning
users.md	Employees, roles, permissions, authentication
equipment.md	Machines, vehicles, repairs, time registration
working-times.md	Time tracking, absences, wage types, logs
reports.md	Daily reports, images, measured quantities
files.md	File uploads, file links, storage
customers.md	Tenant management, customer config, modules
webhooks.md	Webhook dispatching, retry logic
imports.md	Client software imports (cross-reference with module)
gps.md	GPS tracking, geofencing
notifications.md	Email, push, private messages, Slack
dashboard.md	Dashboard widgets, statistics

Not every project has all domains. Only create files for domains that actually exist.

Domain file template

Each domain file should contain:

# Domain: {Name}

## Purpose

{What this domain does in 2-3 sentences}

## Models

| Model | Table | Connection | Key Relationships |
|---|---|---|---|

## Services

| Service | Purpose | Key Methods |
|---|---|---|

## Controllers (API Endpoints)

| Endpoint | Controller | Request | Resource |
|---|---|---|---|

## Jobs & Events

| Class | Type | Trigger | What it does |
|---|---|---|---|

## Business Rules

- {Rule 1: e.g. "A project can only be deleted if it has no working times"}
- {Rule 2}

## Data Flow

{Describe how data moves through this domain — from input to storage to output}

## Dependencies

- Depends on: {other domains}
- Depended on by: {other domains}

Module analysis files

Each module gets its own file in agents/evidence/analysis/modules/. Format:

# Module: {Name}

## Purpose

{What this module does}

## Structure

{Directory tree with key files}

## Public API

{What other parts of the app use from this module: Services, Events, Models}

## Internal Components

{Controllers, Jobs, Commands, Listeners that are module-internal}

## Configuration

{Config files, .env variables, feature flags}

## Testing

{Test suites, test data, stubs}

Detection checklist

Framework & language (multi-stack)

Check	How to detect
PHP runtime + version	composer.json → require.php
Laravel application	artisan file at repo root + laravel/framework in composer.json
Symfony application	bin/console + symfony/framework-bundle in composer.json
Composer package	composer.json without artisan / bin/console
Node.js runtime	package.json exists
TypeScript	tsconfig.json exists
Frontend framework	package.json → react, vue, svelte, solid, astro, @angular/core
Meta-framework	package.json → next, nuxt, remix, sveltekit, astro
Python runtime	pyproject.toml, requirements.txt, setup.py, or Pipfile
Python framework	pyproject.toml / requirements.txt → django, fastapi, flask
Go module	go.mod exists
Rust crate / workspace	Cargo.toml exists
Ruby app	Gemfile → rails, sinatra
.NET project	*.csproj, *.fsproj, or global.json
Java / Kotlin	pom.xml, build.gradle, or build.gradle.kts

After detecting any match, record the stack in the analysis output and select the matching project-analysis-* sub-skill (Laravel, Symfony, Next.js, React, Node/Express, Zend/Laminas) — fall back to project-analysis-core if no framework-specific sub-skill applies.

Project type

Signal	Type
artisan + laravel/framework	Laravel application
bin/console + symfony/framework-bundle	Symfony application
composer.json without artisan / bin/console	Composer package or legacy PHP
package.json with next / nuxt / remix / sveltekit / astro	Meta-framework SSR/SSG app
package.json with express / fastify / koa / hapi	Node HTTP service
package.json with @nestjs/core	NestJS application
pyproject.toml with django / fastapi / flask	Python web app
go.mod with gin-gonic/gin / labstack/echo / gofiber/fiber	Go HTTP service
Module system (app/Modules/, src/modules/, packages/*)	Modular monolith / monorepo
Multi-tenant signal (customer_database, tenant middleware, RLS)	Multi-tenant SaaS
apps/* + packages/* + turbo.json / nx.json / pnpm-workspace	Monorepo

Legacy indicators (stack-aware)

Signal	Meaning
PHP: no declare(strict_types=1) in most files	Pre-modern PHP style
PHP: no typed properties / return types	Legacy PHP (< 7.4)
PHP: no phpstan.neon / rector.php	No static analysis
TS: // @ts-ignore / // @ts-nocheck density; any widespread	Untyped TypeScript
TS: no tsconfig.json strict: true	Loose TypeScript
JS: no ESLint config or eslint.config.*	No linting
Python: no type hints in most signatures; no py.typed	Untyped Python
Python: no mypy.ini / pyrightconfig.json / ruff.toml	No static analysis
Go: no golangci.yml	No lint pipeline
Rust: no clippy.toml and warnings ignored	No lint hygiene
var_dump() / console.log() / print() / fmt.Println() left in code	Legacy debugging patterns
No tests or very few tests	Low test coverage
Mixed naming conventions across the same module	Inconsistent standards

Build & tooling (stack-agnostic)

Check	How to detect
Task runner	Makefile, Taskfile.yml, justfile, package.json scripts, composer.json scripts
Docker	docker-compose.yml, compose.yaml, Dockerfile
CI/CD	.github/workflows/, .gitlab-ci.yml, .circleci/config.yml, azure-pipelines.yml
Quality tools	PHP: phpstan.neon, ecs.php, rector.php. TS/JS: eslint.config.*, .prettierrc*, tsconfig.json. Python: ruff.toml, mypy.ini. Go: .golangci.yml. Rust: clippy.toml
Editor config	.editorconfig
Code review	CODEOWNERS, PR templates (.github/pull_request_template.md)
Dependencies	Lockfile presence: composer.lock, package-lock.json, pnpm-lock.yaml, yarn.lock, poetry.lock, uv.lock, go.sum, Cargo.lock

Integration with other skills

Skill	How it's used
project-docs	Read existing docs before analyzing each area
module-management	Detect and inventory modules
context-create	Create/update context documents
feature-planning	Identify planned but undocumented features
agent-docs-writing	Audit and maintain agent documentation
roadmap-management	Review roadmap status
api-endpoint	Understand endpoint structure for API analysis
database	Understand schema and multi-tenancy for data layer analysis

Workflow

1. Ask scope: Full analysis or specific area (e.g. only domains, only API)?
2. Run phases incrementally — show findings after each phase, ask before continuing.
3. Write files after each phase — don't batch all writing to the end.
4. Ask before creating each file with numbered options:

   > 1. Create — {filename}
   > 2. Skip
   

5. Update existing files if re-running analysis — don't create duplicates.

Output format

1. Structured analysis document in agents/evidence/analysis/
2. Tech stack inventory with versions and dependencies
3. Architecture diagram or module map

Gotcha

• Full project analysis can take several minutes — warn the user about the time investment.
• Don't analyze parts of the codebase that the user hasn't asked about — respect scope.
• Analysis documents go in agents/evidence/analysis/, not in .augment/.

Do NOT

• Do NOT create analysis files without asking — always confirm each creation.
• Do NOT modify existing code — this is analysis only.
• Do NOT commit or push.
• Do NOT overwhelm the user — present findings incrementally, one phase at a time.
• Do NOT analyze third-party code in vendor/ or node_modules/.
• Do NOT duplicate content that already exists in agents/reference/docs/ or agents/settings/contexts/ — reference it instead. Analysis files complement existing docs, they don't replace them.