grepai-ignore-patterns（grepai 忽略模式）

GrepAI Ignore Patterns

This skill covers how to configure ignore patterns to exclude files and directories from GrepAI indexing.

When to Use This Skill

• Excluding test files from search results
• Ignoring generated or vendored code
• Reducing index size by excluding unnecessary files
• Customizing which files GrepAI indexes

How Ignore Patterns Work

GrepAI uses two sources for ignore patterns:

1. .grepai/config.yaml - Custom patterns you define
2. .gitignore - Automatically respected

Configuration Location

# .grepai/config.yaml
ignore:
  - pattern1
  - pattern2

Pattern Syntax

Directory Patterns

ignore:
  # Exact directory name (matches anywhere)
  - node_modules
  - vendor
  - __pycache__

  # With trailing slash (explicit directory)
  - dist/
  - build/
  - coverage/

File Patterns

ignore:
  # Exact filename
  - package-lock.json
  - yarn.lock

  # Wildcard patterns
  - "*.min.js"
  - "*.min.css"
  - "*.map"
  - "*.lock"

Path Patterns

ignore:
  # Paths containing substring
  - /tests/
  - /spec/
  - /__tests__/

  # Specific paths
  - src/generated/
  - api/swagger/

Glob Patterns

ignore:
  # Double star (recursive)
  - "**/test/**"
  - "**/mock/**"

  # Single star (single level)
  - "*.test.js"
  - "*.spec.ts"
  - "*_test.go"

Default Ignore Patterns

GrepAI's default configuration includes:

ignore:
  # Version control
  - .git
  - .svn
  - .hg

  # GrepAI itself
  - .grepai

  # Package managers
  - node_modules
  - vendor
  - .npm
  - .yarn

  # Build outputs
  - target
  - dist
  - build
  - out

  # Cache directories
  - __pycache__
  - .pytest_cache
  - .mypy_cache
  - .cache

  # Framework outputs
  - .next
  - .nuxt
  - .output

Common Ignore Configurations

JavaScript/TypeScript Project

ignore:
  - node_modules
  - dist
  - build
  - coverage
  - .nyc_output
  - "*.min.js"
  - "*.bundle.js"
  - "*.map"
  - package-lock.json
  - yarn.lock
  - pnpm-lock.yaml

Go Project

ignore:
  - vendor
  - bin
  - "*.pb.go"       # Protobuf generated
  - "*_mock.go"     # Mocks
  - "mocks/"
  - go.sum

Python Project

ignore:
  - __pycache__
  - .pytest_cache
  - .mypy_cache
  - .venv
  - venv
  - env
  - "*.pyc"
  - "*.pyo"
  - .eggs
  - "*.egg-info"
  - dist
  - build

Rust Project

ignore:
  - target
  - Cargo.lock
  - "*.rlib"

Java/Kotlin Project

ignore:
  - target
  - build
  - .gradle
  - "*.class"
  - "*.jar"
  - "*.war"

Monorepo

ignore:
  # Common
  - node_modules
  - dist
  - build
  - coverage

  # Specific packages to exclude
  - packages/legacy/
  - packages/deprecated/

  # Generated
  - "**/generated/**"
  - "**/__generated__/**"

Excluding Test Files

To focus search on production code:

ignore:
  # Test directories
  - tests/
  - test/
  - __tests__/
  - spec/

  # Test files by pattern
  - "*.test.js"
  - "*.test.ts"
  - "*.spec.js"
  - "*.spec.ts"
  - "*_test.go"
  - "test_*.py"
  - "*_test.py"

Alternative: Use search boosting instead to penalize (not exclude) tests:

search:
  boost:
    penalties:
      - pattern: /tests/
        factor: 0.5
      - pattern: _test.
        factor: 0.5

Excluding Generated Code

ignore:
  # Generated markers
  - "**/generated/**"
  - "*.generated.*"
  - "*.gen.*"

  # Specific generators
  - "*.pb.go"           # Protobuf
  - "*.graphql.ts"      # GraphQL codegen
  - "*.d.ts"            # TypeScript declarations
  - "swagger_*.go"      # Swagger
  - "openapi_*.ts"      # OpenAPI

Excluding Documentation

ignore:
  - docs/
  - documentation/
  - "*.md"
  - "*.mdx"
  - "*.rst"

Verifying Ignore Patterns

Check what's being indexed:

# Check index status
grepai status

# Output shows file count
# If too high, add more ignore patterns

Common Issues

❌ Problem: Index is too large ✅ Solution: Add more ignore patterns for dependencies and generated files

❌ Problem: Search returns vendor/test code ✅ Solution: Either ignore or use boosting penalties

❌ Problem: Pattern not working ✅ Solution: Check syntax - use quotes for patterns with special characters:

ignore:
  - "*.min.js"  # Correct
  - *.min.js    # May cause YAML parsing issues

❌ Problem: Need to include previously ignored files ✅ Solution: Remove from ignore list and re-run grepai watch

Best Practices

1. Start with defaults: Add patterns as needed
2. Exclude dependencies: Always ignore node_modules, vendor, etc.
3. Exclude build outputs: dist, build, target
4. Exclude lock files: Large, not useful for search
5. Consider boosting vs ignoring: Penalize instead of exclude for test files
6. Quote special characters: "*.min.js" not *.min.js

Re-indexing After Changes

After modifying ignore patterns:

# Stop existing daemon
grepai watch --stop

# Clear index and restart
rm .grepai/index.gob
grepai watch

Output Format

After configuring ignore patterns:

✅ Ignore Patterns Configured

   Patterns: 15 configured

   Categories:
   - Directories: node_modules, vendor, dist, build
   - File types: *.min.js, *.map, *.lock
   - Paths: /tests/, /docs/

   Also respecting: .gitignore

   Run 'grepai watch' to re-index with new patterns.