This is a comment left during a code review.
Path: docs/help/gpt54-codex-agentic-parity.md
Line: 128-135

Comment:
**"Before vs after" table not updated to cover the full 10-PR program**

The column header still reads `After PR A-D` and the five rows only cover PRs A–D. A reader landing on this table after reading the expanded intro gets an incomplete picture — the second-wave improvements from PRs E, H, J, K, and L (auto-activation, doubled scenario pack, tool-call assertions, offline dual-provider mock, run metadata) are silently absent. At minimum the column header should be updated; ideally five new rows should be added.

```suggestion
| Before this program                                                                            | After PRs A–L                                                                             |
| ---------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------- |
| GPT-5.4 could stop after a reasonable plan without taking the next tool step                   | PR A turns "plan only" into "act now or surface a blocked state"                          |
| Strict tool schemas could reject parameter-free or OpenAI/Codex-shaped tools in confusing ways | PR C makes provider-owned tool registration and invocation more predictable               |
| `/elevated full` guidance could be vague or wrong in blocked runtimes                          | PR B gives GPT-5.4 and the user truthful runtime and permission hints                    |
| Replay or compaction failures could feel like the task silently disappeared                    | PR C surfaces paused, blocked, abandoned, and replay-invalid outcomes explicitly          |
| "GPT-5.4 feels worse than Opus" was mostly anecdotal                                           | PR D turns that into the same scenario pack, the same metrics, and a hard pass/fail gate  |
| Strict-agentic had to be opted in per-agent                                                    | PR H auto-activates the contract for all unconfigured GPT-5 / OpenAI-Codex runs          |
| Parity pack covered only five scenarios                                                        | PR E doubles the pack to ten, and any required-scenario fail on either side fails the gate|
| Scenario assertions accepted prose without a real tool call                                    | PR J adds `/debug/requests` tool-call gating on tool-mediated scenarios                  |
| Baseline parity required live Anthropic credentials                                            | PR K routes both providers through the same mock dispatcher — gate runs fully offline     |
| Parity consumers had to trust file paths to identify which summary came from which provider    | PR L records `run.primaryProvider` / `primaryModel` / `providerMode` in each summary     |
```

How can I resolve this? If you propose a fix, please make it concise.

---

This is a comment left during a code review.
Path: docs/help/gpt54-codex-agentic-parity.md
Line: 174-186

Comment:
**Diagram cycles leave the response destination implicit**

`H --> B` and `H --> C` route "Scenario-specific response" back into the two POST-endpoint nodes rather than back to the QA suite runner (`A`). Both B and C are labeled as request paths, so a reader following the arrows left-to-right loses track of where the response ultimately lands. Adding the return arrows directly to `A` (or labeling the back-edges) makes the round-trip explicit:

```
    H -- "OpenAI response" --> A
    H -- "Anthropic response" --> A
```

Minor diagram clarity — does not affect correctness of the other docs.

How can I resolve this? If you propose a fix, please make it concise.

---

This is a comment left during a code review.
Path: docs/help/gpt54-codex-agentic-parity-maintainers.md
Line: 135-143

Comment:
**PR M section is missing a "Does not own" block**

Every other PR section (A through L) has matching "Owns" / "Does not own" bullets. PR M only has "Owns". For consistency and so reviewers can quickly verify scope boundaries, a brief "Does not own" block would complete the pattern, e.g.:

```markdown
Does not own:

- runtime behavior changes
- test coverage
- CI configuration
```

How can I resolve this? If you propose a fix, please make it concise.