| Field | Value |
|---|---|
| Task ID | ADD-XXXX |
| Iteration | Rn (e.g., R1, R2, R3) |
| Date | YYYY-MM-DD |
| Reviewer |
What went wrong? Be specific and factual.
Which ADD phase is the primary source of this failure?
- SCOPE — Task was too broad, ambiguous, or poorly defined.
- FRAME — Context was incomplete, noisy, or contradictory.
- CONSTRAIN — Constraints were vague, missing, or conflicting.
- EXECUTE — Agent underperformed despite adequate preparation.
Why did that phase fail? Be specific.
| Factor | Assessment |
|---|---|
| Was the intent clear? | Yes / No — |
| Were non-goals listed? | Yes / No — |
| Was context complete? | Yes / No — |
| Was context curated (no noise)? | Yes / No — |
| Were constraints testable? | Yes / No — |
| Were forbidden actions listed? | Yes / No — |
| Was the prompt well-composed? | Yes / No — |
Describe the specific failure in the agent's output.
Describe what the human could have done differently in preparation.
Specific changes to make before the next iteration.
| Phase to Revisit | Change |
|---|---|
| SCOPE / FRAME / CONSTRAIN | [Specific change] |
After applying the correction, re-enter the ADD cycle at:
- SCOPE (intent needs revision)
- FRAME (context needs updating)
- CONSTRAIN (constraints need tightening)
- EXECUTE (re-run with updated artifacts)
Is this a recurring failure type?
- First occurrence
- Recurring — seen in tasks: [list Task IDs]
If recurring, should this become:
- A new default constraint in templates
- An anti-pattern entry
- A project-specific rule
- A context pack default inclusion
Additional observations.