audit-env-variables（审计环境变量）

1. Discovery - Find all env files and code references
2. Usage Analysis - Identify which variables are used vs unused
3. Service Detection - Recognize services (Stripe, AWS, Supabase, etc.) and their permission implications
4. Code Path Mapping - Document where each variable is used in the codebase
5. Report Generation - Output a structured markdown document
6. Cleanup (optional) - Safely remove unused variables with user confirmation
7. Regression Prevention - Validate no regressions via build/test validation with automatic rollback

<quick_start> Audit only (default):

1. Scan for .env* files in project root
2. Grep codebase for process.env., import.meta.env., and destructured env patterns
3. Cross-reference declared vs used variables
4. Identify services by naming patterns and categorize permissions
5. Generate markdown report using the template

With cleanup (--cleanup flag): 6. Check for dynamic access patterns that might hide usage 7. Present unused variables and get user confirmation 8. Create backups, remove confirmed variables 9. Run build/tests to detect regressions 10. Auto-rollback if regressions detected </quick_start>

Find all env-related files:

find . -maxdepth 2 -name ".env*" -o -name "env.d.ts" | grep -v node_modules

Common files:

• .env - Local development
• .env.local - Local overrides
• .env.development / .env.production - Environment-specific
• .env.example - Template for required variables

Step 2: Extract Declared Variables

Parse each env file for variable declarations:

Grep pattern: ^[A-Z][A-Z0-9_]+=

Build a list of all declared variables with their source file.

Step 3: Find Code References

Search for environment variable usage patterns:

Direct access:

process.env.VARIABLE_NAME
import.meta.env.VARIABLE_NAME
process.env["VARIABLE_NAME"]

Destructured patterns:

const { API_KEY, DATABASE_URL } = process.env

Framework-specific:

// Next.js public vars
NEXT_PUBLIC_*

// Vite
VITE_*

Use Grep tool with patterns:

process\.env\.([A-Z][A-Z0-9_]+)
import\.meta\.env\.([A-Z][A-Z0-9_]+)

Step 4: Cross-Reference Usage

For each declared variable:

• Used: Found in code references
• Unused: Declared but no code references found
• Undeclared: Referenced in code but not in any env file

Flag potential issues:

• Unused variables (cleanup candidates)
• Undeclared variables (missing from.env.example)
• Variables only in.env but not.env.example (documentation gap)

Step 5: Detect Services and Infer Permissions

Match variable names against known service patterns. See references/service-patterns.md for the complete list.

Categories:

• Database - Connection strings, credentials
• Authentication - JWT secrets, OAuth credentials
• Payment - Stripe, payment processor keys
• Cloud Services - AWS, GCP, Azure credentials
• Third-party APIs - Various service integrations
• Feature Flags - Toggle configurations
• Application Config - URLs, ports, modes

Permission levels:

• Critical - Full account access, billing, admin operations
• High - Read/write access to user data
• Medium - Limited API access, specific operations
• Low - Public keys, non-sensitive configuration

Step 6: Map Code Paths

For each used variable, document:

• File path where it's used
• Function/component context
• Purpose (inferred from surrounding code)

Example:

STRIPE_SECRET_KEY
├── src/lib/stripe.ts:15 - Stripe client initialization
├── src/api/webhooks/stripe.ts:8 - Webhook signature verification
└── src/api/checkout/route.ts:23 - Create checkout session

Step 7: Generate Report

Use the template in templates/env-audit-report.md to generate the final document.

Output to: ENV_AUDIT.md in project root (or user-specified location)

Step 8: Cleanup Unused Variables (Optional)

Trigger: User passes --cleanup flag or explicitly requests cleanup after reviewing the audit report.

8.1 Present Cleanup Candidates

Display unused variables with context:

UNUSED VARIABLES (candidates for removal):

1. OLD_API_KEY (.env, .env.local)
   - Last modified: [file date]
   - No code references found

2. DEPRECATED_SERVICE_URL (.env)
   - Last modified: [file date]
   - No code references found

8.2 Dynamic Access Check

Before confirming removal, search for dynamic access patterns that grep may have missed:

// These patterns indicate variables might be used dynamically:
process.env[variableName]        // Dynamic key access
process.env[`${prefix}_KEY`]     // Template literal access
Object.keys(process.env)         // Iteration over all env vars
{ ...process.env }               // Spread operator

Use Grep with patterns:

process\.env\[
Object\.keys\(process\.env\)
Object\.entries\(process\.env\)
\.\.\.process\.env

If dynamic access patterns found: Flag affected variables for manual review and warn user.

8.3 User Confirmation

Use AskUserQuestion to confirm each removal:

The following variables appear unused. Select which to remove:

[ ] OLD_API_KEY - Remove from .env, .env.local
[ ] DEPRECATED_SERVICE_URL - Remove from .env
[ ] Skip cleanup

⚠️ Variables will be backed up before removal.

Safety rules:

• NEVER auto-delete without explicit user confirmation
• NEVER remove variables from .env.example (it serves as documentation)
• Create backup before any modifications

8.4 Create Backups

Before modifying any file:

# Create timestamped backup directory
mkdir -p .env-backups/$(date +%Y%m%d_%H%M%S)

# Backup each env file being modified
cp .env .env-backups/$(date +%Y%m%d_%H%M%S)/.env.backup
cp .env.local .env-backups/$(date +%Y%m%d_%H%M%S)/.env.local.backup

8.5 Remove Confirmed Variables

For each confirmed variable:

1. Use Edit tool to remove the line from each env file
2. Preserve comments associated with the variable (line immediately above if it starts with #)
3. Log the removal

8.6 Document Cleanup

Append cleanup summary to ENV_AUDIT.md:

## Cleanup Log

**Date:** [timestamp]
**Backup Location:** .env-backups/[timestamp]/

### Removed Variables
| Variable | Removed From | Reason |
|----------|--------------|--------|
| OLD_API_KEY | .env, .env.local | Unused - no code references |

### Preserved (Manual Review Required)
| Variable | Reason |
|----------|--------|
| DYNAMIC_VAR | Dynamic access pattern detected |

Step 9: Regression Prevention

Purpose: Validate that removing unused variables doesn't break the application.

9.1 Pre-Cleanup Baseline

Before removing any variables, capture baseline state:

# Check if project builds successfully
npm run build 2>&1 | tee .env-backups/pre-cleanup-build.log
echo $? > .env-backups/pre-cleanup-build-status

# Run tests if available
npm test 2>&1 | tee .env-backups/pre-cleanup-test.log
echo $? > .env-backups/pre-cleanup-test-status

Store exit codes:

• 0 = Success (baseline is green)
• Non-zero = Already failing (warn user but proceed if they confirm)

9.2 Dynamic Access Detection

Search for patterns that indicate runtime env var access:

High-risk patterns (require manual review):

// Config objects that spread env
const config = { ...process.env }

// Dynamic key construction
const key = `${SERVICE}_API_KEY`
process.env[key]

// Iteration patterns
Object.entries(process.env).filter(([k]) => k.startsWith('FEATURE_'))

// External config loaders
require('dotenv').config({ path: customPath })

Detection commands:

Grep: process\.env\[(?!['"][A-Z])
Grep: Object\.(keys|values|entries)\(process\.env\)
Grep: \.\.\.process\.env
Grep: dotenv.*config

If detected: List affected files and require explicit user acknowledgment before proceeding.

9.3 Post-Cleanup Validation

After removing variables, run validation:

# Verify build still succeeds
npm run build 2>&1 | tee .env-backups/post-cleanup-build.log
POST_BUILD_STATUS=$?

# Verify tests still pass
npm test 2>&1 | tee .env-backups/post-cleanup-test.log
POST_TEST_STATUS=$?

9.4 Regression Detection

Compare pre and post states:

If regression detected:

1. Immediately alert user
2. Offer automatic rollback: ⚠️ REGRESSION DETECTED Build/tests failed after removing variables. Options: 1. Rollback all changes (restore from backup) 2. Rollback specific variables 3. Investigate failure (show diff) 4. Proceed anyway (not recommended)

9.5 Automatic Rollback

If user requests rollback:

# Restore from backup
cp .env-backups/[timestamp]/.env.backup .env
cp .env-backups/[timestamp]/.env.local.backup .env.local

# Verify restoration
npm run build && npm test

9.6 Git Integration (Optional)

If in a git repository, offer branch-based cleanup:

# Create cleanup branch
git checkout -b env-cleanup/$(date +%Y%m%d)

# After cleanup, changes can be reviewed via PR
git add .env .env.local
git commit -m "chore: remove unused environment variables

Removed:
- OLD_API_KEY (unused)
- DEPRECATED_SERVICE_URL (unused)

Backup: .env-backups/[timestamp]/"

Benefits:

• Easy rollback via git checkout -
• Changes visible in PR review
• CI/CD will validate before merge

<reference_index> Service Patterns: references/service-patterns.md - Known services and their variable naming conventions </reference_index>

<success_criteria> Audit is complete when:

• All.env files discovered and parsed
• All code references found
• Variables categorized as used/unused/undeclared
• Services detected and permission levels assigned
• Code paths documented for each variable
• Markdown report generated with all sections
• Report saved to specified location

Cleanup is complete when (if --cleanup requested):

• Dynamic access patterns checked
• User confirmed variables to remove
• Backups created in.env-backups/
• Variables removed from env files (not.env.example)
• Cleanup log appended to report

Regression prevention is complete when:

• Pre-cleanup build/test baseline captured
• Post-cleanup validation run
• No regressions detected OR rollback performed
• Git branch created (if opted in) </success_criteria>