autoworker:subtask-plan — Complete Verification Plan (Gate 1.3 Step C+D+E)

Invoked after autoworker:subtask-init completes. Fills in upstream traceability table, L1-L4 verification plan, and solution self-check.

Execution Flow

1. Locate and Check Subtask

Glob `subtask_*.md` (exclude subtask_template.md) →
  0 found → stop, prompt to create subtask
  1 found → use directly (backward compatible)
  multiple → grep `status:` to filter:
    - Files without status field treated as active (backward compatible)
    - Exactly 1 active → use it
    - 0 active → list all files + status, prompt user to choose
    - >1 active → report anomaly
→ Read → check first-half completeness

Pre-checks:

• Assumptions table "Result" column has empty rows → FAIL, prompt to complete autoworker:subtask-init first
• User confirmation record is empty → FAIL
• Goal/success criteria is empty → FAIL

2. Fill In Upstream Verification Traceability Table

Source: Read plan file (if exists) + review conversation for user's verification/testing requirements.

Edit subtask's "Upstream Verification Traceability" table:

Rules:

• Copy verbatim from upstream text — not summarize (the act of copying forces you to see what upstream required)
• When no upstream requirements exist, write one line: No upstream requirements | Self-designed | —
• Subsequent L1-L4 are not allowed to be lower than the traceability table; downgrades must state delta rationale

2.5. Silent Failure Analysis

Read the subtask's "Acceptance Criteria" table, for each metric ask:

If this metric deviates, which L1-L4 layer would catch it?

• Metric has no corresponding test → add in subsequent L2-L4
• All metrics have corresponding tests → continue

Purpose: Prevent acceptance criteria defined in the plan but not covered by the verification plan (silent failure).

3. Write L1-L4 Verification Plan

Cross-reference the traceability table, Edit subtask to fill in each layer's verification plan:

L1 Build:

• Specific build/check commands → expected output

L2 Unit (can skip but state reason):

• Specific function calls + expected input/output

L3 Chain (can skip but state reason):

• Commands feeding downstream with actual upstream output

L4 End-to-End (mandatory):

• Complete steps simulating actual user operation path
• Operation table: # | Action | Expected result | Actual result (leave empty)

Verification coverage table: | Modified file | Change content | Corresponding verification item | Each file planned for modification must have a corresponding verification item.

Constraints:

• Cannot be lower than upstream requirements in the traceability table. Downgrades must be explicitly noted in the delta rationale column
• Each metric in the acceptance criteria table must have at least one corresponding test in L2-L4

4. Acceptance Criteria Coverage Check

Build metric-to-test mapping table:

• Uncovered metrics found → return to Step 3 and add corresponding tests
• All covered → continue

5. Solution Self-Check (Four Questions)

Answer each one (any unsatisfactory → report the issue, do not continue):

0. Is the direction right? Derived from problem essence, or pattern-matched from similar scenarios?
1. Is the abstraction level right? Does the caller naturally possess this capability?
2. 10x robustness: Still robust if input diversity increases 10x?
3. Does a simpler solution exist? Same effect, less code/assumptions?

Edit self-check results into subtask's progress log.

6. Output

Subtask ready: subtask_<sequence>_<name>.md
- Traceability: <N> upstream requirements aligned
- Verification plan: L1 done, L2 <done/skip>, L3 <done/skip>, L4 done
- Coverage: <N>/<N> files have verification items
- Acceptance criteria coverage: <N>/<N> metrics have corresponding tests
- Self-check: 4/4 passed
→ Automatically invoking autoworker:dispatch

7. Chain: Immediately Invoke autoworker:dispatch

After outputting the summary above, immediately invoke autoworker:dispatch. Do not wait for user instructions, do nothing else.

Important Notes

• Traceability table is an anti-downgrade mechanism: Skipping the traceability table and writing L1-L4 directly = may miss upstream requirements
• L4 is mandatory: End-to-end verification cannot be skipped — it's the ultimate safeguard for the user path
• Complete coverage: Uncovered modified files → add verification items before continuing
• Chaining is mandatory: Must invoke autoworker:dispatch after completion, cannot skip or manually substitute