Tasks: test and verification as nodes in the PDLC

The last spec artifact. Derived from the locked design.md and testing-plan.md; each _Test:_ names a matrix row of the plan.

Task list

• [x] 1. Bundle the testing-plan.md template
  • New skills/the-loop/templates/testing-plan.md with the four gated sections (Test matrix, Verification environment, Evidence plan, Verification results), the type catalogue, the activities checklist, and the two security rules (credentials by reference; redact evidence).
  • Depends on: none
  • Requirements: R1.4, R2.1, R2.2, R4.2, R5.1
  • Test: T1 — pytest cli/tests/test_graph_parity.py -k p3 (red→green: P3 fails while the template is absent or short a section)
• [x] 2. Add the two nodes to the shipped graph
  • test-planning (produces testing-plan.md, locked + four sections) between design-approval and tasks-breakdown; verification (re-produces testing-plan.md, checkmarks: complete + results section) between implementation and self-review; edges rerouted.
  • Depends on: 1
  • Requirements: R1.1, R1.2, R3.1, R3.2
  • Test: T1 — pytest cli/tests/test_graph_model.py (red→green)
• [x] 3. Track the new artifacts in the manifest
  • docs/specs/<id>/testing-plan.md @ test-planning; docs/specs/<id>/evidence/ as an optional directory.
  • Depends on: 2
  • Requirements: R4.2, R6.2
  • Test: T1 — pytest cli/tests/test_graph_parity.py -k "p1 or p2" (red→green: P2 fails while the graph gates an untracked name)
• [x] 4. Declare the phases in the schema and both configs
  • workflow.phases enum + default; this repo's .the-loop/harness-config.yaml and the bundled templates/harness-config.yaml; tokenEconomy stage defaults for test-planning and verification in schema and both configs.
  • Depends on: 2
  • Requirements: R6.1, R6.4
  • Test: T1 — pytest cli/tests/test_graph_parity.py -k p4 (red→green)
• [x] 5. Unit-cover the two nodes, including the not-a-skip regression
  • cli/tests/test_graph_model.py: both nodes present with phases and stages, edges rerouted, and verification declares produces so its gate cannot silently skip.
  • Depends on: 2
  • Requirements: R3.1, R3.2
  • Test: T1 — pytest cli/tests/test_graph_model.py
• [x] 6. Integration-cover the verification gate end to end
  • New cli/tests/test_graph_verification_integration.py driving the real hook chain over a temporary spec folder: unticked activity blocks, empty results blocks, complete plan passes. Gherkin docstrings, Requirement: links.
  • Depends on: 2, 1
  • Requirements: R3.2, R3.3
  • Test: T2 — pytest cli/tests/test_graph_verification_integration.py (red→green)
• [x] 7. Render the new sequence in the skill and references
  • SKILL.md (artifact chain, phase sequence, principles, command list), reference/workflow.md (phases, the two nodes, evidence), reference/testing.md (the plan, the catalogue, evidence and redaction, the facilitate-don't-own rule), reference/context.md phase list.
  • Depends on: 2
  • Requirements: R6.3, R4.3, R5.2, R5.3
  • Test: T3 — make lint (markdownlint) + review
• [x] 8. Wire the commands
  • New commands/create-testing-plan.md and commands/verify-work.md; chain them from create-design.md, create-tasks-plan.md, execute-tasks.md and work-on.md; init.md creates the two new labels.
  • Depends on: 2, 7
  • Requirements: R1.1, R3.1, R6.3
  • Test: T3 — make lint + review
• [x] 9. Update the templates the loop authors from
  • design.md (Testing strategy points at the plan), tasks.md (_Test:_ names a matrix row), execution-log.md (phase enum, transition rows).
  • Depends on: 2
  • Requirements: R1.3, R6.3
  • Test: T1 — pytest cli/tests/test_graph_parity.py (P3 covers the templates)
• [x] 10. Fold in the capability docs and the decision record
  • docs/capabilities/testing-and-contracts.md (the plan + verification node), spec-workflow.md, process-graph.md, capabilities.md index; docs/reports/labels-and-dashboards.md; docs/decisions/decision-060.md + index; guide/README/architecture phase sequences.
  • Depends on: 7
  • Requirements: R6.3
  • Test: T3 — make lint + review
• [x] 11. Added during implementation: a skip must not short-circuit a chain
  • Found while writing T6: run_chain stopped at the first non-pass result, including skip, so hooks behind a skipping one never ran and a chain ending in a skip routed on the outcome "skip" — for which no edge is declared. implementation (chain ends in a no-op verify-tests) therefore parked at no_edge, making the implementation → verification edge unreachable. Fixed in graph/chain.py.
  • Depends on: 2
  • Requirements: R3.1
  • Test: T1 — pytest cli/tests/test_graph_chain.py -k Skip; T2 — test_implementation_reaches_verification_rather_than_parking (red→green)
• [x] 13. Added in PR review: one gate for the design and the plan
  • Owner's call on PR #166: move test-planning between design and design-approval so the single human gate approves both artifacts; design-approval records feedback into each; changes-requested returns to design, which re-derives the plan. Spec (D1/D2), decision-060, skill, references, commands and capability docs updated to match.
  • Depends on: 2
  • Requirements: R1.1
  • Test: T1 — pytest cli/tests/test_graph_model.py -k Testing (red→green on the three edge/feedback assertions)
• [x] 12. Dogfood: this work item's own testing plan, executed
  • docs/specs/issue-163/testing-plan.md authored at planning and completed at verification with results + evidence under docs/specs/issue-163/evidence/.
  • Depends on: 1–11, 13
  • Requirements: R4.1, R4.2, R4.5
  • Test: T1, T2, T3 — the full suite, recorded as evidence

Dependency graph (DAG)

1 → 2 → 3
      → 4
      → 5
      → 6 (also ← 1)
      → 11
      → 13
      → 7 → 8
      → 9
        7 → 10
1..11, 13 → 12

Checkpoints

After tasks 2–6 (the executable surface) run make test; after 7–10 run make lint. Task 12 is the verification node itself: the full suite plus lint, recorded in the plan's Verification results with committed evidence, before the review chain runs.

Review comments

Appended by the-loop's record-feedback hook when a human gate approves with comments (issue-109).