fix: align benchmark agent settings and enforce MiniSWE output budget

[Yofuria]

Aligns benchmark recipes with the audited target/mutation agent settings and makes MINISWE_MAX_OUTPUT_LIMIT an enforced runtime override in the candidate adapter.

Validation:

• 63 targeted adapter/recipe tests passed
• repository-wide Ruff, format, and ty checks passed
• end-to-end runtime canaries recorded max_output_tokens=10000, cost_limit=3.0, and reasoning.effort=high

Summary by CodeRabbit

• New Features

  • Added support for configuring the maximum model output length through the MINISWE_MAX_OUTPUT_LIMIT environment setting.
  • OpenAI-based models now use the configured limit, with a default maximum applied when no setting is provided.

• Bug Fixes

  • Invalid limits, including zero, negative, or non-numeric values, are now rejected with a clear validation error.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Team

Run ID: 027732de-f698-4ba4-8cb5-cd521c52dce7

📥 Commits

Reviewing files that changed from the base of the PR and between f046c39 and 4930da4.

📒 Files selected for processing (2)

• src/evolve/integrations/harbor/miniswe_candidate.py
• tests/test_miniswe_harbor_wrapper.py

Included review availability: Your plan provides up to 1 included review per hour; 0 remain after this review.

---

📝 Walkthrough

Walkthrough

The MiniSWE wrapper now forwards MINISWE_MAX_OUTPUT_LIMIT, validates it as a positive integer, and applies it to OpenAI models. Tests cover forwarding, runtime overrides, and invalid values.

Changes

MiniSWE output limit

Layer / File(s)	Summary
Forward the runtime limit src/evolve/integrations/harbor/miniswe_candidate.py, tests/test_miniswe_harbor_wrapper.py	The source environment forwards MINISWE_MAX_OUTPUT_LIMIT. Tests verify the forwarded value.
Validate and apply the limit src/evolve/integrations/harbor/miniswe_candidate.py, tests/test_miniswe_harbor_wrapper.py	The model setup parses positive integer values, applies them to OpenAI models, and retains the 64,000 default when unset. Tests cover overrides and invalid values.

Estimated code review effort: 2 (Simple) | ~10 minutes

Merge Risk: 🔵 Low · up to 4930d

The change enforces a runtime output-token override for MiniSWE models, but the setting has no explicit maximum, so a misconfigured value could increase execution time or cost beyond the intended budget. The PR is mergeable with owner awareness or follow-up to document the trusted configuration owner and bound the allowed value.

Suggested reviewers: zimuwang-real, yesiyu403

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 8 functions across 2 files.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly identifies the two main changes: aligning benchmark agent settings and enforcing the MiniSWE output budget.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

• Fix all pre-merge checks with AI

✨ Finishing Touches 💡 1

📝 Generate docstrings 💡

• Create stacked PR
• Commit on current branch

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}