File Filtering

Smart Defaults

promptext automatically excludes common non-source files:

• Dependencies: node_modules/, vendor/, bower_components/
• Build Output: dist/, build/, out/, target/
• Version Control: .git/, .svn/, .hg/
• IDE Files: .idea/, .vscode/, *.sublime-*
• Cache: __pycache__/, .sass-cache/, .npm/
• Logs: logs/, *.log, tmp/

Disable default rules:

promptext -u=false  # Only use explicit excludes

Multi-Layered Lock File Detection

promptext automatically excludes package lock files using a sophisticated 3-layer detection system, saving 50K-100K+ tokens per project:

Layer 1: Signature-Based Detection (99% confidence)

Pattern matching with content verification for 15+ lock file formats:

• Node.js: package-lock.json, yarn.lock, pnpm-lock.yaml, bun.lockb, .pnp.cjs
• Python: poetry.lock, Pipfile.lock, pdm.lock
• PHP: composer.lock
• Ruby: Gemfile.lock
• Rust: Cargo.lock
• Go: go.sum
• .NET: packages.lock.json, project.assets.json
• Java: gradle.lockfile

Requires multiple signature matches to avoid false positives.

Layer 2: Ecosystem-Aware Detection (95% confidence)

Automatically detects package managers by scanning for manifest files:

• Detects package.json → excludes Node.js lock files
• Detects go.mod → excludes go.sum
• Detects pyproject.toml → excludes Python lock files
• Context-aware for uncommon lock file formats

Layer 3: Generated File Detection (85% confidence)

Heuristic detection for auto-generated files:

• Checks for markers: @generated, autogenerated, do not edit
• Low-entropy pattern analysis for repetitive structures
• Catches minified files, bundles, source maps (.min.js, .map, etc.)

Token Savings Examples

Project Type	Lock File	Size	Token Savings
Node.js (React)	package-lock.json	500KB	~80,000 tokens
Python (Poetry)	poetry.lock	200KB	~60,000 tokens
Rust	Cargo.lock	150KB	~45,000 tokens
PHP (Laravel)	composer.lock	400KB	~70,000 tokens

Binary Detection

Automatically skips binary files by detecting:

• Null bytes in first 512 bytes
• Invalid UTF-8 encoding

No need to specify binary extensions — images, executables, and archives are automatically excluded.

Custom Patterns

Pattern Types

Directory matching:

excludes:
  - test/           # Any 'test' directory
  - internal/tmp/   # Specific path

Wildcards:

excludes:
  - "*.test.go"     # Test files
  - ".aider*"       # Generated files

Exact matches:

excludes:
  - config.json     # Specific file
  - src/constants.go # Exact path

Configuration

Config file:

.promptext.yml

excludes:
  - test/
  - "*.generated.*"
  - docs/private/

Command line:

promptext -x "test/,*.generated.*,docs/private/"

GitIgnore Integration

Automatically respects .gitignore patterns. Disable with:

promptext -g=false

Filter Priority

1. Default patterns (if enabled)
2. GitIgnore patterns (if enabled)
3. Custom excludes (config + command line)

All patterns are combined and deduplicated for optimal performance.