Ignore Patterns¶

ContextHarness supports a .contextignore file for excluding files and directories from context scanning. This is useful for large projects with directories that don't need to be analyzed.

Overview¶

The .contextignore file uses the same syntax as .gitignore. When you run ch init, an empty .contextignore template is created in your project root.

Default Patterns¶

ContextHarness automatically ignores common directories even without a .contextignore file:

Category	Patterns
Version Control	`.git/`, `.svn/`, `.hg/`
Dependencies	`node_modules/`, `vendor/`, `.venv/`, `venv/`, `__pycache__/`
Build Artifacts	`dist/`, `build/`, `.next/`, `.nuxt/`, `out/`, `target/`
Cache	`.pytest_cache/`, `.mypy_cache/`, `.ruff_cache/`, `.cache/`
IDE	`.idea/`, `.vscode/`
Lock Files	`package-lock.json`, `yarn.lock`, `pnpm-lock.yaml`, `poetry.lock`, `uv.lock`

Syntax¶

The .contextignore file follows .gitignore syntax:

# Comments start with #
node_modules/          # Directory pattern (trailing slash)
*.pyc                  # Wildcard pattern
!important.pyc         # Negation (un-ignore)
**/test_*.py           # Recursive wildcard
apps/legacy/           # Specific path

Pattern Types¶

Pattern	Matches
`foo/`	Directory named `foo` and all contents
`*.log`	All files ending in `.log`
`!keep.log`	Exclude `keep.log` from being ignored
`**/temp/`	Any `temp/` directory at any depth
`docs/*.md`	Markdown files directly in `docs/`

Examples¶

Monorepo with Multiple Apps¶

Exclude apps you're not working on:

# Ignore legacy app
apps/legacy-app/

# Ignore deprecated packages
packages/deprecated/
packages/old-utils/

# But keep the shared package
!packages/shared/

Large Generated Content¶

Exclude generated files that inflate context:

# Generated code
.generated/
codegen/
*.generated.ts

# API clients
src/api/generated/

Test Data¶

Exclude large test fixtures:

# Large test fixtures
tests/fixtures/large/
test-data/
__fixtures__/

# But keep small fixtures
!tests/fixtures/small/

Documentation Sites¶

Exclude built documentation:

# Built docs
docs/_build/
site/
.docusaurus/

Vendored Dependencies¶

Exclude third-party code copied into repo:

vendor/
third_party/
external/

Programmatic Usage¶

The IgnoreService can be used programmatically in Python:

from context_harness.services import IgnoreService
from pathlib import Path

# Load ignore configuration
service = IgnoreService()
config = service.load_or_default(Path("."))

# Check if a path should be ignored
if service.should_ignore(Path("node_modules/foo.js"), config):
    print("Ignoring node_modules/foo.js")

# Filter a list of paths
paths = [Path("src/main.py"), Path("node_modules/x.js")]
filtered = service.filter_paths(paths, config)
# Returns: [Path("src/main.py")]

# Get patterns for external tools
result = service.get_exclusion_patterns()
if isinstance(result, Success):
    for pattern in result.value:
        print(f"--exclude={pattern}")

How It Works¶

On ch init: A minimal .contextignore template is created
On /baseline: The discovery phase respects ignore patterns
Pattern matching: Uses the pathspec library for accurate gitignore-style matching
Layered patterns: Default patterns are applied first, then your custom patterns

Tips¶

Start minimal: Only add patterns when you notice context overload
Use negation sparingly: ! patterns can be confusing; prefer explicit includes
Test patterns: Run /baseline and check what's included in PROJECT-CONTEXT.md
Share patterns: Commit .contextignore so your team benefits