ship-safe-fix

Ship Safe — Auto-Fix Security Issues

You are using Ship Safe's remediation engine to automatically fix security issues in this project.

Step 1: Preview the fixes

Always start with a dry run to show what will change:

npx ship-safe@latest remediate $ARGUMENTS --dry-run 2>&1

If $ARGUMENTS is empty, default to . --all (fix both secrets and agent findings):

npx ship-safe@latest remediate . --all --dry-run 2>&1

Flags:

• No --all flag → only fixes hardcoded secrets (moves to env vars)
• With --all → also fixes: TLS bypass (rejectUnauthorized: false), Docker :latest tags, debug mode enabled, dangerouslySetInnerHTML without sanitization, shell: true in exec/spawn

Step 2: Present the preview

Show the user what will be changed:

1. List each file that will be modified
2. Show the before/after for each change
3. Group by fix type (secrets, TLS, Docker, debug, XSS, shell)
4. Note any files that will be created (.env.example, .env)

Step 3: Apply fixes (with confirmation)

Ask the user if they want to proceed. If yes:

npx ship-safe@latest remediate . --all --yes 2>&1

If the user only wants to fix secrets (not agent findings):

npx ship-safe@latest remediate . --yes 2>&1

Step 4: Post-fix verification

After applying fixes:

1. Run a quick scan to verify secrets were removed:

   npx ship-safe@latest scan . --json 2>/dev/null
   

2. Report the results — how many issues were fixed vs. remaining

3. For remaining issues that couldn't be auto-fixed, offer to fix them manually by reading the code and applying targeted changes

Step 5: Follow-up actions

Suggest:

• Review .env.example — make sure variable names make sense
• Add .env to .gitignore if not already there
• Rotate exposed secrets — run npx ship-safe rotate . for step-by-step guides
• Update baseline — run /ship-safe-baseline . to update after fixes
• Stage changes — offer to stage the modified files with git

Important Notes

• Ship Safe creates a backup before modifying files — the user can revert if something breaks
• The --all flag is important for fixing agent-level findings beyond just secrets
• Never display actual secret values, even in the dry run output
• If a fix might break functionality (e.g., shell: false in exec), warn the user to test
• For dangerouslySetInnerHTML fixes, note that DOMPurify needs to be installed: npm install dompurify