Review Mode (Core)

[dedrisian-oai]

📝 Review Mode -- Core

This PR introduces the Core implementation for Review mode:

• New op Op::Review { prompt: String }: spawns a child review task with isolated context, a review‑specific system prompt, and a Config.review_model.
• EnteredReviewMode: emitted when the child review session starts. Every event from this point onwards reflects the review session.
• ExitedReviewMode(Option<ReviewOutputEvent>): emitted when the review finishes or is interrupted, with optional structured findings:

{
  "findings": [
    {
      "title": "<≤ 80 chars, imperative>",
      "body": "<valid Markdown explaining *why* this is a problem; cite files/lines/functions>",
      "confidence_score": <float 0.0-1.0>,
      "priority": <int 0-3>,
      "code_location": {
        "absolute_file_path": "<file path>",
        "line_range": {"start": <int>, "end": <int>}
      }
    }
  ],
  "overall_correctness": "patch is correct" | "patch is incorrect",
  "overall_explanation": "<1-3 sentence explanation justifying the overall_correctness verdict>",
  "overall_confidence_score": <float 0.0-1.0>
}

Questions

Why separate out its own message history?

We want the review thread to match the training of our review models as much as possible -- that means using a custom prompt, removing user instructions, and starting a clean chat history.

We also want to make sure the review thread doesn't leak into the parent thread.

Why do this as a mode, vs. sub-agents?

1. We want review to be a synchronous task, so it's fine for now to do a bespoke implementation.
2. We're still unclear about the final structure for sub-agents. We'd prefer to land this quickly and then refactor into sub-agents without rushing that implementation.

[chatgpt-codex-connector]

Codex Review: Here are some suggestions.

Reply with @codex fix comments to fix any unresolved comments.

About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you open a pull request for review, mark a draft as ready, or comment "@codex review". If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex fix this CI failure" or "@codex address that feedback".

[jif-oai]

A lot of not necessary clone()
Handling the memory in Rust is not easy but we should not take the habit to clone everything as it can have a large impact on performances at large scale

[pakrym-oai]

Few comments. The main one (but non-blocking) is around removing is_review_mode from the session and encapsulating it in the Task::review() overload.

[zemaj]

This is a great feature! Can't wait to see how it performs.

It looks like the review thread only gets custom instructions + whatever input the /review is called with.

If /review is used without any input would it make sense to give the thread a curated view of recent activity? i.e. recent user instructions and perhaps just the patches applied or recent tool calls? Or does this not make sense given how the review models are trained?

[pakrym-oai]

Overall layering looks good, please address @jif-oai's comments before merging.

[jif-oai]

Fix the pipeline but LGTM 👍