| name | description |
|---|---|
task-direction-approval |
Use when considering switching libraries/tools, changing architecture, replacing automation with manual workarounds, adding new dependencies, or changing approach while keeping the same library - explains root cause, offers 2-3 options with trade-offs, and requests explicit user choice |
This skill prevents unauthorized direction changes by forcing explicit user consent before deviating from the original requirements.
- Switching from the original tech/library to alternatives
- Replacing an automated approach with a manual workaround
- Changing architecture or design patterns
- Delivering different results than requested
- "direction change", "can't use X", "let's switch to Y", "workaround", "manual", "different approach", "alternative library", "rewrite", "change architecture"
❌ Wrong Response:
"Code generator X failed, so I'll define types manually instead."
✅ Correct Response:
"Code generator X failed due to authentication error. Options available:
- Add authentication headers
- Try different endpoint
- Download schema/spec file directly Which approach would you prefer?"
- Explain the failure root cause clearly (what failed, why it failed, evidence).
- Present 2-3 viable options with trade-offs (speed, risk, maintenance, correctness).
- Ask for explicit user choice.
- Only proceed after approval.
- If user rejects all options: Present the constraints clearly and ask the user to define the new direction — do not proceed unilaterally.
- Fix obvious typos
- Correct configuration mistakes
- User says "any method is fine"
- Use
investigation-modeskill when there are repeated failures or the situation is unclear.