learn-project

Learn Project

Scan the current project's source code to detect patterns, classify tooling, and propose domain rules. Unlike /forge domain extract (which reads dotforge's internal memory), this skill reads the CODE directly.

Step 1: Detect dependency files

Read whichever exist (skip missing):

• package.json — extract dependencies + devDependencies keys
• pyproject.toml — extract [project.dependencies] and [tool.*] sections
• requirements.txt — read all lines
• go.mod — extract require block
• Podfile — read all lines
• Gemfile — read all lines
• pom.xml / build.gradle / build.gradle.kts — extract dependency declarations

Collect a flat list of dependency names.

Step 2: Scan imports

Find the top 20 most-imported libraries across the codebase:

# Python
grep -rh "^import \|^from " --include="*.py" . 2>/dev/null | sed 's/import //;s/from //;s/ .*//' | sort | uniq -c | sort -rn | head -20

# TypeScript/JavaScript
grep -rh "from ['\"]" --include="*.ts" --include="*.tsx" --include="*.js" --include="*.jsx" . 2>/dev/null | sed "s/.*from ['\"]//;s/['\"].*//" | grep -v '^\.' | sort | uniq -c | sort -rn | head -20

# Go
grep -rh "\"" --include="*.go" . 2>/dev/null | grep -E '^\s+"' | sed 's/.*"//;s/".*//' | sort | uniq -c | sort -rn | head -20

# Swift
grep -rh "^import " --include="*.swift" . 2>/dev/null | sed 's/import //' | sort | uniq -c | sort -rn | head -20

Step 3: Scan project structure

# Directory structure (2 levels deep, ignore hidden/vendor)
find . -maxdepth 2 -type d ! -path '*/\.*' ! -path '*/node_modules*' ! -path '*/.venv*' ! -path '*/vendor*' ! -path '*/__pycache__*' | sort

# Config files present
ls -1 .eslintrc* .prettierrc* ruff.toml pyproject.toml tsconfig.json biome.json jest.config* vitest.config* vite.config* webpack.config* Makefile Dockerfile docker-compose* .env.example 2>/dev/null

Step 4: Classify patterns

From Steps 1-3, classify into categories. Only report categories with HIGH CONFIDENCE (direct evidence in code):