Enforce human requirement ledgers in PR follow-ups

[wwind123]

Fixes #559

Implements the approved plan for auditable human-requirement handling in post-plan-approval PR coder follow-ups.

• Requires and validates the top-level disposition ledger.
• Preserves direct-discussion fallback acknowledgement semantics.
• Threads fallback context through repair and renders evidenced ledgers in public comments.
• Updates regression coverage and structured fixtures.

[wwind123]

{
"schema_version": 1,
"kind": "coder_followup",
"state": "blocking",
"summary": "Implemented the approved human-requirement ledger flow for PR coder follow-ups and opened PR #567.",
"addressed_items": [],
"remaining_items": [],
"human_requirement_dispositions": [],
"human_requirements": {
"addressed_ids": [],
"checked_discussion_directly": false
},
"tests_run": [
"python3 -m pytest tests/test_response_validation.py tests/test_protocol.py tests/test_prompts.py tests/test_repair.py tests/test_comment_rendering.py tests/test_human_requirement_dispositions.py",
"python3 -m pytest"
]
}

-- OpenAI Codex: gpt-5.6-luna

[wwind123]

Review verdict: Blocking

The PR-review coder-followup flow correctly gains a per-requirement human_requirement_dispositions ledger (validated with duplicate/unknown/missing checks, rendered in public comments, wired through the repair pass) reusing the machinery from the plan-review flow. Logic and new/updated tests look sound, but the documented coder_followup JSON schema in README.md and docs/local_agent_loop.md was not updated for the new required field, and a couple of touched test/prompt files carry mechanical-edit artifacts.

Blocking issues

• README.md (~lines 592-597) and docs/local_agent_loop.md (~lines 1100-1119) still show the coder_followup structured JSON example/description without the new required human_requirement_dispositions field. Following the documented example as-is would now fail schema validation (the field is required per _expect_exact_keys in protocol.py). Update both docs to include the field and describe the addressed/blocked/not-applicable + evidence ledger.

Same-PR follow-ups

• tests/test_protocol.py has several dict literals where "human_requirement_dispositions": [] was accidentally duplicated as a key (e.g. lines 2101-2102, 2164-2165, 2191-2192, 2226-2227), plus inconsistent indentation from the mechanical edit (e.g. lines 2134-2136, 2360-2362, 2386-2388, 2412-2414); tests/test_orchestrator_pr.py has similar misindentation around lines 3309-3319. Please clean these up.
• In _coder_followup_human_requirements_instruction (src/coding_review_agent_loop/repair.py, ~lines 1316-1318), the trailing sentence "When the list is (none) without direct-discussion acknowledgement, both ledgers must be empty." is appended unconditionally, including in the branch where surfaced requirement IDs are non-empty, which makes the repair prompt reference a (none) case that doesn't apply there. Scope that sentence to the actually-empty-list branch.

-- Anthropic Claude: claude-sonnet-5

[wwind123]

Review verdict: Blocking

The PR correctly implements human requirement ledgers for coder follow-ups, but introduces several duplicate dictionary keys and formatting issues in the tests that should be fixed before merge.

Same-PR follow-ups

• Remove duplicated "human_requirement_dispositions": [], keys in the test payloads in tests/test_protocol.py (e.g. test_validate_structured_coder_followup_accepts_v1_payload, test_validate_structured_coder_followup_rejects_note_for_unlisted_item, etc.).
• Fix incorrect indentation for the newly added human_requirement_dispositions keys in tests/test_protocol.py (e.g. lines ~2135, 2361, 2387, 2413).
• Fix incorrect indentation for the human_requirement_dispositions list comprehension in structured_coder_followup inside tests/agent_loop_helpers.py.
• Consider renaming test_non_plan_prompts_do_not_request_plan_disposition_json in tests/test_prompts.py since prompts[0] and prompts[1] now explicitly request human_requirement_dispositions.

-- Google Antigravity: Gemini 3.1 Pro (High)

[wwind123]

Coder follow-up

Addressed all requested documentation, fixture, indentation, repair-prompt, and test-name follow-up findings.

Addressed items

• item-1: Blocking issue from Anthropic Claude: claude-sonnet-5, round 1: README.md (~lines 592-597) and docs/local_agent_loop.md (~lines 1100-1119) still show the `coder_...
  • Resolution: Updated README.md and docs/local_agent_loop.md with the required human_requirement_dispositions field and its per-requirement evidence ledger rules.
• item-2: Same-PR follow-up from Anthropic Claude: claude-sonnet-5, round 1: tests/test_protocol.py has several dict literals where "human_requirement_dispositions": [] was...
  • Resolution: Removed duplicated payload keys and corrected inconsistent test fixture indentation in tests/test_protocol.py.
• item-3: Same-PR follow-up from Anthropic Claude: claude-sonnet-5, round 1: In _coder_followup_human_requirements_instruction (src/coding_review_agent_loop/repair.py, ~lin...
  • Resolution: Scoped the empty-list ledger sentence in _coder_followup_human_requirements_instruction to the actually empty surfaced-ID branch.
• item-4: Same-PR follow-up from Google Antigravity: Gemini 3.1 Pro (High), round 1: Remove duplicated "human_requirement_dispositions": [], keys in the test payloads in `tests/tes...
  • Resolution: Removed all duplicated human_requirement_dispositions keys from the affected protocol test payloads.
• item-5: Same-PR follow-up from Google Antigravity: Gemini 3.1 Pro (High), round 1: Fix incorrect indentation for the newly added human_requirement_dispositions keys in `tests/tes...
  • Resolution: Corrected indentation of the affected human_requirement_dispositions and related fields in tests/test_protocol.py.
• item-6: Same-PR follow-up from Google Antigravity: Gemini 3.1 Pro (High), round 1: Fix incorrect indentation for the human_requirement_dispositions list comprehension in `structu...
  • Resolution: Corrected indentation of the structured_coder_followup disposition list comprehension in tests/agent_loop_helpers.py.
• item-7: Same-PR follow-up from Google Antigravity: Gemini 3.1 Pro (High), round 1: Consider renaming test_non_plan_prompts_do_not_request_plan_disposition_json in `tests/test_pro...
  • Resolution: Renamed the prompt test to describe its human-requirement disposition behavior accurately.

Remaining items

• None.

Tests run

• python3 -m pytest tests/test_protocol.py tests/test_repair.py tests/test_prompts.py tests/test_orchestrator_pr.py tests/test_agent_loop.py
• python3 -m pytest

-- OpenAI Codex: gpt-5.6-luna

[wwind123]

Review verdict: Approved

All round-1 findings are resolved: README.md and docs/local_agent_loop.md now document the required human_requirement_dispositions ledger for coder follow-ups, the repair.py prompt wording only mentions the (none) case when the list is actually empty, and the duplicate-key/indentation mechanical-edit artifacts in tests/test_protocol.py, tests/test_orchestrator_pr.py, and tests/agent_loop_helpers.py are cleaned up, with tests/test_prompts.py renamed to reflect its updated assertions. No new issues found in this round's diff.

Prior unresolved item dispositions

• [item-1] Blocking issue from Anthropic Claude: claude-sonnet-5, round 1: README.md (~lines 592-597) and docs/local_agent_loop.md (~lines 1100-1119) still show the `coder_... -> resolved
• [item-2] Same-PR follow-up from Anthropic Claude: claude-sonnet-5, round 1: tests/test_protocol.py has several dict literals where "human_requirement_dispositions": [] was... -> resolved
• [item-3] Same-PR follow-up from Anthropic Claude: claude-sonnet-5, round 1: In _coder_followup_human_requirements_instruction (src/coding_review_agent_loop/repair.py, ~lin... -> resolved
• [item-4] Same-PR follow-up from Google Antigravity: Gemini 3.1 Pro (High), round 1: Remove duplicated "human_requirement_dispositions": [], keys in the test payloads in `tests/tes... -> resolved
• [item-5] Same-PR follow-up from Google Antigravity: Gemini 3.1 Pro (High), round 1: Fix incorrect indentation for the newly added human_requirement_dispositions keys in `tests/tes... -> resolved
• [item-6] Same-PR follow-up from Google Antigravity: Gemini 3.1 Pro (High), round 1: Fix incorrect indentation for the human_requirement_dispositions list comprehension in `structu... -> resolved
• [item-7] Same-PR follow-up from Google Antigravity: Gemini 3.1 Pro (High), round 1: Consider renaming test_non_plan_prompts_do_not_request_plan_disposition_json in `tests/test_pro... -> resolved

-- Anthropic Claude: claude-sonnet-5

[wwind123]

Review verdict: Approved

All prior issues regarding human_requirement_dispositions documentation, schema validation tests, duplication, indentation, prompt instructions, and test renaming have been successfully addressed.

Prior unresolved item dispositions

• [item-1] Blocking issue from Anthropic Claude: claude-sonnet-5, round 1: README.md (~lines 592-597) and docs/local_agent_loop.md (~lines 1100-1119) still show the `coder_... -> resolved
• [item-2] Same-PR follow-up from Anthropic Claude: claude-sonnet-5, round 1: tests/test_protocol.py has several dict literals where "human_requirement_dispositions": [] was... -> resolved
• [item-3] Same-PR follow-up from Anthropic Claude: claude-sonnet-5, round 1: In _coder_followup_human_requirements_instruction (src/coding_review_agent_loop/repair.py, ~lin... -> resolved
• [item-4] Same-PR follow-up from Google Antigravity: Gemini 3.1 Pro (High), round 1: Remove duplicated "human_requirement_dispositions": [], keys in the test payloads in `tests/tes... -> resolved
• [item-5] Same-PR follow-up from Google Antigravity: Gemini 3.1 Pro (High), round 1: Fix incorrect indentation for the newly added human_requirement_dispositions keys in `tests/tes... -> resolved
• [item-6] Same-PR follow-up from Google Antigravity: Gemini 3.1 Pro (High), round 1: Fix incorrect indentation for the human_requirement_dispositions list comprehension in `structu... -> resolved
• [item-7] Same-PR follow-up from Google Antigravity: Gemini 3.1 Pro (High), round 1: Consider renaming test_non_plan_prompts_do_not_request_plan_disposition_json in `tests/test_pro... -> resolved

-- Google Antigravity: Gemini 3.1 Pro (High)