Skip to main content

ICW-E4S08T07-Test-Design: FR-011 Multi-Epic Task Splitting

Task: E4:S08:T07
Related FR: FR-011
Host Task: T07-task-splitting-across-multiple-canonical-epics-fr-011.md
Created: 2026-04-09

Test Strategy​

Validate split-candidate detection, task-level assignment, explainability output, ID remap determinism, and reference rewrite integrity for FR-011.

Coverage Matrix​

  1. Multi-epic candidate detection (FR-1 / AC-1)

    • Input: source epic with two or more partial canonical matches.
    • Expectation: split-candidate classification is emitted.

  2. Task-content assignment analysis (FR-2 / AC-2)

    • Input: mixed task corpus with versioning and workflow tasks.
    • Expectation: tasks route to canonical epic targets aligned to content signals.

  3. Split execution behavior (FR-3 / AC-3)

    • Input: eligible split-candidate epic with N tasks.
    • Expectation: all tasks assigned, no task omitted or multiply assigned.

  4. Split-plan explainability output (FR-4 / AC-4)

    • Input: generated split plan.
    • Expectation: per-task rationale, confidence/tie-break metadata, and mapping table.

  5. Task ID remap correctness (FR-5 / AC-5)

    • Input: split output + target story/task spaces.
    • Expectation: IDs remapped deterministically and collision-safe.

  6. Reference rewrite integrity (FR-6 / AC-6)

    • Input: docs containing pre-split IDs.
    • Expectation: structured references updated; unresolved ambiguous references reported.

Failure Modes​

  • Single-best-match fallback still used despite multi-match eligibility.
  • Non-deterministic assignment when scores are tied.
  • Rationale omitted or inconsistent with assignment decisions.
  • ID collision or unstable numbering between runs.
  • Partial reference updates leaving stale IDs in linked docs.

Red/Green Verification Sequence​

  1. Start with failing assertions for FR-1..FR-6 behaviors.
  2. Implement detection and assignment logic until behavioral tests pass.
  3. Implement explainability schema checks.
  4. Implement deterministic remap and replay-idempotence checks.
  5. Implement reference rewrite checks and unresolved-report checks.
  6. Re-run full regression set on mixed synthetic and realistic fixtures.

Regression Guard Checklist​

  • Deterministic replay test: identical input produces identical split mapping.
  • Tie-break policy test: explicit deterministic order for equal-confidence tasks.
  • Orphan guard test: no task dropped from split outputs.
  • Reference guard test: stale IDs absent from structured targets post-rewrite.
  • Explainability guard test: each assignment has rationale payload.

Evidence for Handoff​

  • Test case list mapped to FR-011 FR/AC identifiers.
  • Fixture set definitions for mixed-responsibility epic scenarios.
  • Pass/fail checklist for deterministic and integrity guards.