chore: PDD sync for #1384

[prompt-driven-github]

Summary

Add pdd/prompts/capability_ir_python.prompt — the prompt spec for a new capability_ir module that compiles <capabilities> tag text into a language-neutral effect IR (EffectItem / ContractEffectIR dataclasses) using only stdlib re, with no AST or code-analysis imports.

Closes #1384

Changes Made

Prompts Modified

• pdd/prompts/capability_ir_python.prompt (NEW) — prompt spec defining EffectItem, ContractEffectIR, and parse_capabilities_ir(text) entry point; specifies modal normalization (MAY / MUST NOT → MUST_NOT), resource normalization algorithm (lowercase → strip . → split on first list separator → snake_case), sentinel return for missing/empty tag, and stdlib-only constraint.

Documentation Updated

None — capability_ir is an internal parsing leaf module; <capabilities> syntax is already documented in docs/prompting_guide.md.

Review Checklist

• Prompt syntax is valid
• PDD conventions followed
• Documentation is up to date

Next Steps After Merge

1. Regenerate code from modified prompts in dependency order:

   pdd sync capability_ir

2. Run tests to verify functionality:

   pytest -vv tests/test_capability_ir.py

3. Deploy if applicable

---

Created by pdd change workflow

[prompt-driven-github]

Step 7/8: Review Loop Final Report

PR: #1388
Issue: #1384
issue_aligned: false
active-reviewer: codex
reviewer-status: codex=findings claude=fixer fresh-final=missing
fresh-final-review: missing
verified-head-sha: none
remote-pr-head-sha: none
max-rounds-reached: false
max-cost-reached: false
max-duration-reached: false

Summary

Fixer claude could not address codex's findings.

Per-Reviewer Status

Reviewer	Status
codex	findings
claude	fixer
fresh-final	missing

Findings

Severity	Status	Location	Finding	Required fix	Reviewer
medium	open	pdd/capability_ir.py:75	The parser misses the acceptance example's resource normalization for plural send targets and provider-secret compound phrases, so the foundational IR does not match the documented capabilities contract users are already shown in docs/prompting_guide.md.	Update the source prompt normalization rule, generated parser, example, and tests so the exact issue sample parses to email and secrets as specified. Add a regression test for the full sample block from the issue to prevent the PR's narrower examples from masking this mismatch.	codex
low	open	architecture.json:10172	The architecture entry no longer matches the prompt metadata source of truth. A sync-architecture run would rewrite this field, leaving the PR with generated metadata drift before merge.	Regenerate or manually align the architecture.json reason with the XML-safe prompt reason, using Parses capabilities tag text into a target-neutral effect IR (EffectItem / ContractEffectIR).	codex

Fixer Rationale

• pdd/capability_ir.py:75: The parser misses the acceptance example's resource normalization for plural send targets and provider-secret compound phrases, so the foundational IR does not match the documented capabilities contract users are already shown in docs/prompting_guide.md. (fixer=claude fixer_summary='All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":429,"duration_ms":581,"duration_api_ms":0,"num_turns":1,"result":"You've hit your limit · resets 9:30pm (UTC)","stop_reason":"stop_sequence","session_id":"644026b4-0534-4377-acd9-3ccd662afc66","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"server_tool_use":{"web_search_requests":0,"web_fetch_reques'; verification=unverified)
• architecture.json:10172: The architecture entry no longer matches the prompt metadata source of truth. A sync-architecture run would rewrite this field, leaving the PR with generated metadata drift before merge. (fixer=claude fixer_summary='All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":429,"duration_ms":581,"duration_api_ms":0,"num_turns":1,"result":"You've hit your limit · resets 9:30pm (UTC)","stop_reason":"stop_sequence","session_id":"644026b4-0534-4377-acd9-3ccd662afc66","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"server_tool_use":{"web_search_requests":0,"web_fetch_reques'; verification=unverified)

Fixes Attempted

• round=1 fixer=claude fixer_result=failed push_status=not_attempted local_sha=none pushed_sha=none changed_files=none verification=unverified summary=All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":429,"duration_ms":581,"duration_api_ms":0,"num_turns":1,"result":"You've hit your limit · resets 9:30pm (UTC)","stop_reason":"stop_sequence","session_id":"644026b4-0534-4377-acd9-3ccd662afc66","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"server_tool_use":{"web_search_requests":0,"web_fetch_reques

[prompt-driven-github]

Step 7/8: Review Loop Final Report

PR: #1388
Issue: #1384
issue_aligned: false
active-reviewer: codex
reviewer-status: codex=findings claude=fixer fresh-final=missing
fresh-final-review: missing
verified-head-sha: none
remote-pr-head-sha: none
max-rounds-reached: false
max-cost-reached: false
max-duration-reached: false

Summary

Fixer claude could not address codex's findings.

Per-Reviewer Status

Reviewer	Status
codex	findings
claude	fixer
fresh-final	missing

Findings

Severity	Status	Location	Finding	Required fix	Reviewer
medium	open	pdd/capability_ir.py:75	The parser misses the documented acceptance example's resource normalization for plural send targets and provider-secret compound phrases. This means the foundational IR users will get from the canonical capabilities examples does not match the issue contract.	Update the source prompt normalization rule, generated parser, example, and tests so the exact issue/docs sample parses to resource='email' and resource='secrets' as specified. Add a regression test for the full sample block from the issue/docs so narrower examples do not mask this mismatch.	codex
low	open	architecture.json:10172	The architecture entry no longer matches the prompt metadata source of truth. A sync-architecture run would rewrite this field, leaving the PR with generated metadata drift before merge.	Regenerate or manually align architecture.json with the XML-safe prompt reason: 'Parses capabilities tag text into a target-neutral effect IR (EffectItem / ContractEffectIR).'	codex
low	open	pdd/capability_ir.py:25	The generated Python file contains trailing whitespace, causing the repository readiness check to fail.	Strip trailing whitespace from pdd/capability_ir.py and keep the generated artifact clean before merge.	codex

Fixer Rationale

• pdd/capability_ir.py:75: The parser misses the documented acceptance example's resource normalization for plural send targets and provider-secret compound phrases. This means the foundational IR users will get from the canonical capabilities examples does not match the issue contract. (fixer=claude fixer_summary='All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":403,"duration_ms":717,"duration_api_ms":0,"num_turns":1,"result":"Your organization does not have access to Claude. Please login again or contact your administrator.","stop_reason":"stop_sequence","session_id":"55b4da02-4304-4ed1-b8c2-f7b6ccf9fa39","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"ser'; verification=unverified)
• architecture.json:10172: The architecture entry no longer matches the prompt metadata source of truth. A sync-architecture run would rewrite this field, leaving the PR with generated metadata drift before merge. (fixer=claude fixer_summary='All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":403,"duration_ms":717,"duration_api_ms":0,"num_turns":1,"result":"Your organization does not have access to Claude. Please login again or contact your administrator.","stop_reason":"stop_sequence","session_id":"55b4da02-4304-4ed1-b8c2-f7b6ccf9fa39","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"ser'; verification=unverified)
• pdd/capability_ir.py:25: The generated Python file contains trailing whitespace, causing the repository readiness check to fail. (fixer=claude fixer_summary='All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":403,"duration_ms":717,"duration_api_ms":0,"num_turns":1,"result":"Your organization does not have access to Claude. Please login again or contact your administrator.","stop_reason":"stop_sequence","session_id":"55b4da02-4304-4ed1-b8c2-f7b6ccf9fa39","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"ser'; verification=unverified)

Fixes Attempted

• round=1 fixer=claude fixer_result=failed push_status=not_attempted local_sha=none pushed_sha=none changed_files=none verification=unverified summary=All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":403,"duration_ms":717,"duration_api_ms":0,"num_turns":1,"result":"Your organization does not have access to Claude. Please login again or contact your administrator.","stop_reason":"stop_sequence","session_id":"55b4da02-4304-4ed1-b8c2-f7b6ccf9fa39","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"ser

[prompt-driven-github]

Step 7/8: Review Loop Final Report

PR: #1388
Issue: #1384
issue_aligned: false
active-reviewer: codex
reviewer-status: codex=findings claude=fixer fresh-final=missing
fresh-final-review: missing
verified-head-sha: none
remote-pr-head-sha: none
max-rounds-reached: false
max-cost-reached: false
max-duration-reached: false

Summary

Fixer claude could not address codex's findings.

Per-Reviewer Status

Reviewer	Status
codex	findings
claude	fixer
fresh-final	missing

Findings

Severity	Status	Location	Finding	Required fix	Reviewer
medium	open	pdd/capability_ir.py:63	The generated parser does not satisfy the documented acceptance example's resource normalization, so users parsing the canonical capabilities block get different IR than the issue contract promises.	Update the source prompt normalization contract, regenerate pdd/capability_ir.py and context/capability_ir_example.py as needed, and add a regression test using the full issue sample that asserts resource='email' and resource='secrets'. The fixer did not address this; current tests pass only because they use narrower examples.	codex
medium	open	pdd/capability_ir.py:31	A full prompt without a tag can be treated as if capabilities are present, violating the missing-tag acceptance criterion and creating false-positive contract effects from ordinary bullet lists.	Clarify and implement the API boundary so full raw prompts without a tag return the false/empty sentinel while still supporting explicitly parsed tag content. Add tests for a no-tag prompt containing MAY/MUST NOT bullets.	codex
low	open	architecture.json:10172	The architecture entry no longer matches the prompt metadata source of truth, so a sync-architecture run would rewrite generated metadata after merge.	Regenerate or manually align architecture.json with the XML-safe prompt reason: "Parses capabilities tag text into a target-neutral effect IR (EffectItem / ContractEffectIR)."	codex
low	open	pdd/capability_ir.py:36	The generated Python artifact contains trailing whitespace, causing the repository readiness check to fail for the committed PR content.	Strip trailing whitespace from pdd/capability_ir.py and ensure regenerated artifacts remain clean before merge.	codex

Fixer Rationale

• pdd/capability_ir.py:63: The generated parser does not satisfy the documented acceptance example's resource normalization, so users parsing the canonical capabilities block get different IR than the issue contract promises. (fixer=claude fixer_summary='All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":429,"duration_ms":552,"duration_api_ms":0,"num_turns":1,"result":"You've hit your limit · resets 9:30pm (UTC)","stop_reason":"stop_sequence","session_id":"6604fccf-633a-475f-96a9-ae169cca05c3","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"server_tool_use":{"web_search_requests":0,"web_fetch_reques'; verification=unverified)
• pdd/capability_ir.py:31: A full prompt without a tag can be treated as if capabilities are present, violating the missing-tag acceptance criterion and creating false-positive contract effects from ordinary bullet lists. (fixer=claude fixer_summary='All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":429,"duration_ms":552,"duration_api_ms":0,"num_turns":1,"result":"You've hit your limit · resets 9:30pm (UTC)","stop_reason":"stop_sequence","session_id":"6604fccf-633a-475f-96a9-ae169cca05c3","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"server_tool_use":{"web_search_requests":0,"web_fetch_reques'; verification=unverified)
• architecture.json:10172: The architecture entry no longer matches the prompt metadata source of truth, so a sync-architecture run would rewrite generated metadata after merge. (fixer=claude fixer_summary='All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":429,"duration_ms":552,"duration_api_ms":0,"num_turns":1,"result":"You've hit your limit · resets 9:30pm (UTC)","stop_reason":"stop_sequence","session_id":"6604fccf-633a-475f-96a9-ae169cca05c3","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"server_tool_use":{"web_search_requests":0,"web_fetch_reques'; verification=unverified)
• pdd/capability_ir.py:36: The generated Python artifact contains trailing whitespace, causing the repository readiness check to fail for the committed PR content. (fixer=claude fixer_summary='All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":429,"duration_ms":552,"duration_api_ms":0,"num_turns":1,"result":"You've hit your limit · resets 9:30pm (UTC)","stop_reason":"stop_sequence","session_id":"6604fccf-633a-475f-96a9-ae169cca05c3","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"server_tool_use":{"web_search_requests":0,"web_fetch_reques'; verification=unverified)

Fixes Attempted

• round=1 fixer=claude fixer_result=failed push_status=not_attempted local_sha=none pushed_sha=none changed_files=none verification=unverified summary=All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":429,"duration_ms":552,"duration_api_ms":0,"num_turns":1,"result":"You've hit your limit · resets 9:30pm (UTC)","stop_reason":"stop_sequence","session_id":"6604fccf-633a-475f-96a9-ae169cca05c3","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"server_tool_use":{"web_search_requests":0,"web_fetch_reques

[prompt-driven-github]

Step 7/8: Review Loop Final Report

PR: #1388
Issue: #1384
issue_aligned: false
active-reviewer: codex
reviewer-status: codex=findings claude=fixer fresh-final=missing
fresh-final-review: missing
verified-head-sha: none
remote-pr-head-sha: none
max-rounds-reached: false
max-cost-reached: false
max-duration-reached: false

Summary

Fixer claude could not address codex's findings.

Per-Reviewer Status

Reviewer	Status
codex	findings
claude	fixer
fresh-final	missing

Findings

Severity	Status	Location	Finding	Required fix	Reviewer
medium	open	pdd/capability_ir.py:63	The generated parser does not satisfy the documented acceptance example's resource normalization, so users parsing the canonical capabilities block get different IR than the issue contract promises.	Update the source prompt normalization contract, regenerate pdd/capability_ir.py and context/capability_ir_example.py as needed, and add a regression test using the full issue sample that asserts resource='email' and resource='secrets'.	codex
medium	open	pdd/capability_ir.py:31	A full prompt without a tag can be treated as if capabilities are present, violating the missing-tag acceptance criterion and creating false-positive contract effects from ordinary bullet lists.	Clarify and implement the API boundary so full raw prompts without a tag return the false/empty sentinel while still supporting explicitly parsed tag content. Add tests for a no-tag prompt containing MAY/MUST NOT bullets.	codex
low	open	architecture.json:10172	The architecture entry no longer matches the prompt metadata source of truth, so a sync-architecture run would rewrite generated metadata after merge.	Regenerate or manually align architecture.json with the XML-safe prompt reason: 'Parses capabilities tag text into a target-neutral effect IR (EffectItem / ContractEffectIR).'	codex
low	open	pdd/capability_ir.py:36	The generated Python artifact contains trailing whitespace, causing the repository readiness check to fail for the committed PR content.	Strip trailing whitespace from pdd/capability_ir.py and ensure regenerated artifacts remain clean before merge.	codex

Fixer Rationale

• pdd/capability_ir.py:63: The generated parser does not satisfy the documented acceptance example's resource normalization, so users parsing the canonical capabilities block get different IR than the issue contract promises. (fixer=claude fixer_summary='All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":403,"duration_ms":451,"duration_api_ms":0,"num_turns":1,"result":"Your organization does not have access to Claude. Please login again or contact your administrator.","stop_reason":"stop_sequence","session_id":"e457a480-08f5-4f49-b280-2667b7fe7c15","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"ser'; verification=unverified)
• pdd/capability_ir.py:31: A full prompt without a tag can be treated as if capabilities are present, violating the missing-tag acceptance criterion and creating false-positive contract effects from ordinary bullet lists. (fixer=claude fixer_summary='All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":403,"duration_ms":451,"duration_api_ms":0,"num_turns":1,"result":"Your organization does not have access to Claude. Please login again or contact your administrator.","stop_reason":"stop_sequence","session_id":"e457a480-08f5-4f49-b280-2667b7fe7c15","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"ser'; verification=unverified)
• architecture.json:10172: The architecture entry no longer matches the prompt metadata source of truth, so a sync-architecture run would rewrite generated metadata after merge. (fixer=claude fixer_summary='All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":403,"duration_ms":451,"duration_api_ms":0,"num_turns":1,"result":"Your organization does not have access to Claude. Please login again or contact your administrator.","stop_reason":"stop_sequence","session_id":"e457a480-08f5-4f49-b280-2667b7fe7c15","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"ser'; verification=unverified)
• pdd/capability_ir.py:36: The generated Python artifact contains trailing whitespace, causing the repository readiness check to fail for the committed PR content. (fixer=claude fixer_summary='All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":403,"duration_ms":451,"duration_api_ms":0,"num_turns":1,"result":"Your organization does not have access to Claude. Please login again or contact your administrator.","stop_reason":"stop_sequence","session_id":"e457a480-08f5-4f49-b280-2667b7fe7c15","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"ser'; verification=unverified)

Fixes Attempted

• round=1 fixer=claude fixer_result=failed push_status=not_attempted local_sha=none pushed_sha=none changed_files=none verification=unverified summary=All agent providers failed: anthropic: Exit code 1: {"type":"result","subtype":"success","is_error":true,"api_error_status":403,"duration_ms":451,"duration_api_ms":0,"num_turns":1,"result":"Your organization does not have access to Claude. Please login again or contact your administrator.","stop_reason":"stop_sequence","session_id":"e457a480-08f5-4f49-b280-2667b7fe7c15","total_cost_usd":0,"usage":{"input_tokens":0,"cache_creation_input_tokens":0,"cache_read_input_tokens":0,"output_tokens":0,"ser