Tasks: ad-hoc tasks that run no PDLC process

The last spec artifact (requirements → design → testing plan → tasks). A DAG of implementation tasks derived from the approved design and testing plan.

Task list

• [x] 1. Name the fourth loop and localize loop resolution

  • cli/the_loop/graph/model.py: add PDLC_ADHOC_LOOP, extend SHIPPED_LOOPS, add OUTER_PATH_LOOPS, LOOP_FOR_CONTROL_COMMAND and resolve_outer_loop; export them.
  • Depends on: none
  • Requirements: R1.1, R2.3, R2.4
  • Test: T1 — pytest tests/test_graph_adhoc.py -k "resolve_outer_loop or control_command" (red→green)

• [x] 2. Ship pdlc-adhoc-loop.yaml

  • Three walkable nodes (work, review, complete) plus terminal cleanup and escalated; no produces, no validate-artifacts, no skipSets, no required/skippable; command: do-task on every node that renders a resume hint.
  • Depends on: 1
  • Requirements: R1.1–R1.6
  • Test: T1 — pytest tests/test_graph_adhoc.py -k "compiles or gates_nothing or no_selection" (red→green)

• [x] 3. Add the classify-adhoc-reply hook

  • New cli/the_loop/graph/hooks/adhoc.py, registered from hooks/__init__.py; reuses feedback._authorized_comments; newest authorized comment decides; outcomes done and more-work; waiting when nothing authorized has arrived.
  • Security-relevant (trust boundary 2, design.md §Security design): the negative tests are the unauthorized-reply and self-authored-"done" cases.
  • Depends on: 1
  • Requirements: R3.1–R3.4
  • Test: T1/T8 — pytest tests/test_graph_adhoc.py -k "reply or self_authored or unauthorized" (red→green)

• [x] 4. Add the do control keyword

  • cli/the_loop/control.py: DO, COMMANDS, _ARMING_COMMANDS, SPAWN_COMMANDS, DEFAULT_KEYWORDS; no parser change.
  • Security-relevant (trust boundary 1): the negative tests are the unauthorized-arming and two-keyword-refusal cases, plus the-loop done not matching.
  • Depends on: none
  • Requirements: R2.1, R2.2
  • Test: T1/T8 — pytest tests/test_graph_adhoc.py -k "keyword or whole_token or refused" (red→green)

• [x] 5. Route the new loop through the four resolution seams

  • graph/bootstrap.py (OUTER_PATH_LOOPS membership), graphlink.py (_outer_loop_name, and the ad-hoc iterate on: line in render_graph_context), core/graphs.py (_recorded_loop).
  • Depends on: 1, 2, 4
  • Requirements: R2.3–R2.5, R4.2, R5.2
  • Test: T1/T10 — pytest tests/test_graph_adhoc.py tests/test_graph_contribution.py tests/test_graphlink.py tests/test_core_graphs.py (red→green)

• [x] 6. Ship /the-loop:do-task

  • commands/do-task.md, modelled on contribute-to.md; states what the loop omits and forbids authoring a spec chain.
  • Depends on: 2
  • Requirements: R4.1, R4.3
  • Test: T1 — pytest tests/test_graph_adhoc.py -k command asserts every node's command: names do-task

• [x] 7. Config surface: schema, template, docs page

  • .the-loop/cli-config.schema.json → copied byte-identically to cli/the_loop/schemas/cli-config.schema.json; skills/the-loop/templates/cli-config.yaml; docs/config/cli/routing-options.md.
  • Depends on: 4
  • Requirements: R2.1
  • Test: T1 — pytest tests/test_config_schema_parity.py tests/test_docs_parity.py

• [x] 8. Documentation and the decision record

  • skills/the-loop/SKILL.md, skills/the-loop/reference/workflow.md, docs/capabilities/process-graph.md, docs/capabilities/webhook-triggers.md, docs/reference/commands.md, docs/decisions/decision-083.md (+ the decisions index).
  • Depends on: 2, 4, 6
  • Requirements: R1, R2, R3, R4, R5
  • Test: T1 — pytest tests/test_docs_parity.py

• [x] 9. Verification

  • Execute testing-plan.md; record results and commit redacted evidence under docs/specs/issue-225/evidence/.
  • Depends on: 1–8
  • Requirements: all
  • Test: T1, T2, T3, T8, T10

Dependency graph (DAG)

flowchart LR
  T1[1 · loop names] --> T2[2 · graph yaml]
  T1 --> T3[3 · classify-adhoc-reply]
  T4[4 · do keyword] --> T5[5 · resolution seams]
  T1 --> T5
  T2 --> T5
  T2 --> T6[6 · do-task command]
  T4 --> T7[7 · schema + docs page]
  T2 --> T8[8 · docs + decision]
  T4 --> T8
  T6 --> T8
  T3 --> T9[9 · verification]
  T5 --> T9
  T7 --> T9
  T8 --> T9

Checkpoints

After tasks 2, 5 and 8: run uv run --project cli pytest and append an execution-log entry. After task 9 the verification results and evidence are committed, then the review phases run.

Review comments