Configuration

Configuration

Customize Drift for your project.

Configuration File

Drift stores configuration in .drift/config.json:

{
  "version": "2.0.0",
  "project": {
    "id": "uuid-generated-on-init",
    "name": "my-project",
    "description": "Optional project description",
    "initializedAt": "2024-01-01T00:00:00.000Z"
  },
  "severity": {},
  "ignore": [
    "node_modules/**",
    "dist/**",
    "**/*.test.ts"
  ],
  "ci": {
    "failOn": "error",
    "reportFormat": "text"
  },
  "learning": {
    "autoApproveThreshold": 0.95,
    "minOccurrences": 3,
    "semanticLearning": true
  },
  "performance": {
    "maxWorkers": 4,
    "cacheEnabled": true,
    "incrementalAnalysis": true,
    "cacheTTL": 3600
  },
  "features": {
    "callGraph": true,
    "boundaries": true,
    "dna": true,
    "contracts": true
  }
}

---

Configuration Options

Project Settings

Option	Type	Default	Description
project.id	string	auto-generated	Unique project identifier (UUID)
project.name	string	folder name	Project display name
project.description	string	-	Optional project description
project.initializedAt	string	auto-set	ISO timestamp of initialization

Ignore Settings

Option	Type	Default	Description
ignore	array	see below	Glob patterns to exclude from scanning
include	array	-	Glob patterns to include (allowlist mode)

Default ignore patterns:

[
  "node_modules/**",
  "dist/**",
  "build/**",
  ".git/**",
  "coverage/**",
  "*.min.js",
  "*.bundle.js",
  "vendor/**",
  "__pycache__/**",
  ".venv/**",
  "target/**",
  "bin/**",
  "obj/**"
]

Allowlist mode: If include is specified, ONLY those paths are scanned. This takes precedence over ignore patterns.

CI Settings

Option	Type	Default	Description
ci.failOn	string	"error"	Severity that causes CI failure: error, warning, info, none
ci.reportFormat	string	"text"	Output format: text, json, sarif, github
ci.uploadResults	boolean	false	Upload results to dashboard

Learning Settings

Option	Type	Default	Description
learning.autoApproveThreshold	number	0.95	Auto-approve patterns above this confidence (0-1)
learning.minOccurrences	number	3	Minimum occurrences before pattern is detected
learning.semanticLearning	boolean	true	Enable semantic learning

Note: learning.autoApproveThreshold is the recommended way to auto-approve patterns. Set to 1.0 to disable auto-approval, or lower (e.g., 0.8) to be more permissive.

Performance Settings

Option	Type	Default	Description
performance.maxWorkers	number	4	Maximum parallel workers
performance.cacheEnabled	boolean	true	Enable caching
performance.incrementalAnalysis	boolean	true	Enable incremental scanning
performance.cacheTTL	number	3600	Cache TTL in seconds

Feature Flags

Option	Type	Default	Description
features.callGraph	boolean	true	Enable call graph analysis
features.boundaries	boolean	true	Enable data boundary scanning
features.dna	boolean	true	Enable styling DNA analysis
features.contracts	boolean	true	Enable API contract verification

---

.driftignore

Exclude files from scanning. Uses the same syntax as .gitignore.

Important: Drift automatically respects your .gitignore patterns. You don't need to duplicate them in .driftignore. The .driftignore file is for additional exclusions specific to Drift scanning.

# .driftignore - Additional exclusions for Drift
# (Your .gitignore patterns are already respected automatically)

# Test files (if you don't want patterns from tests)
*.test.ts
*.spec.ts
__tests__/

# Generated files
*.generated.ts
*.d.ts

# Large data files Drift shouldn't analyze
*.sql
*.csv
fixtures/

How it works:

1. Drift reads both .gitignore and .driftignore
2. Patterns from both files are combined
3. Files matching EITHER pattern are excluded from scanning
4. This is the default behavior — no configuration needed

---

Environment Variables

Variable	Description
DRIFT_CONFIG	Path to config file
DRIFT_CACHE_DIR	Cache directory
DRIFT_LOG_LEVEL	Log level: debug, info, warn, error
DRIFT_NO_COLOR	Disable colored output
DRIFT_PARALLEL	Number of parallel workers

---

Per-Project Configuration

Multiple Projects

Register multiple projects:

drift projects add ~/code/backend --name backend
drift projects add ~/code/frontend --name frontend

Each project has its own .drift/ directory and configuration.

Switching Projects

drift projects switch backend
drift status  # Shows backend status

---

CI Configuration

GitHub Actions

- name: Drift Check
  run: |
    npm install -g driftdetect
    drift init --yes
    drift scan
    drift check --ci --fail-on warning --format github

GitLab CI

drift:
  script:
    - npm install -g driftdetect
    - drift init --yes
    - drift scan
    - drift check --ci --fail-on warning --format gitlab

Pre-commit Hook

# .husky/pre-commit
drift check --staged --fail-on error

---

MCP Server Configuration

Rate Limiting

{
  "mcp": {
    "rateLimit": {
      "global": 100,
      "expensive": 10
    }
  }
}

Caching

{
  "mcp": {
    "cache": {
      "enabled": true,
      "ttl": 300000,
      "maxSize": 100
    }
  }
}

Tool Filtering (Intelligent Defaults)

By default, Drift automatically detects your project's languages and only exposes relevant MCP tools. A TypeScript project won't see Go, Rust, C++, PHP, or Java tools — reducing AI context overhead from 49 tools to ~35.

How it works:

1. On MCP server startup, Drift scans for language markers (package.json, Cargo.toml, go.mod, etc.)
2. Only language-specific tools for detected languages are exposed
3. Core tools (patterns, status, context, etc.) are always available

Override detection:

{
  "mcp": {
    "tools": {
      "languages": ["typescript", "python"],
      "include": ["drift_go"],
      "exclude": ["drift_wpf"],
      "all": false
    }
  }
}

Option	Type	Description
languages	array	Override detected languages
include	array	Additional tools to include
exclude	array	Tools to exclude
all	boolean	Set true to disable filtering (expose all 49 tools)

Supported languages: typescript, javascript, python, java, php, go, rust, cpp, csharp, wpf

---

Sensitive Data Configuration

Custom Sensitive Fields

{
  "boundaries": {
    "sensitiveFields": [
      "ssn",
      "social_security",
      "tax_id",
      "bank_account"
    ]
  }
}

Sensitivity Categories

{
  "boundaries": {
    "categories": {
      "pii": ["email", "phone", "address"],
      "financial": ["credit_card", "bank_account"],
      "health": ["diagnosis", "prescription"],
      "credentials": ["password", "api_key", "token"]
    }
  }
}

---

Resetting Configuration

# Reset to defaults
rm -rf .drift
drift init

# Keep patterns, reset config only
rm .drift/config.json
drift init --yes