Fix blank responses from thinking models

[michaelneale]

Problem

Thinking models can appear to return a blank response in Mesh. When reasoning_format is omitted, Mesh currently parses generated reasoning and then silently discards it. The model is producing tokens and the GPU is working, but clients receive no visible stream while the model is reasoning.

This differs from llama-server/LM Studio and from harness expectations, where the default is to pass model reasoning through.

Fix

Default an omitted reasoning_format to auto instead of Mesh's hidden mode. Reasoning is now streamed to clients as reasoning_content, so the reproduced Qwen3.8 request no longer appears hung or empty.

Explicitly configured formats such as hidden and none keep their existing behavior. This does not change whether a model thinks; it fixes whether generated thinking is silently lost.

Validation

• cargo fmt --all --check
• cargo check -p skippy-server
• cargo clippy -p skippy-server --all-targets -- -D warnings
• cargo check -p mesh-llm
• cargo clippy -p mesh-llm --all-targets -- -D warnings
• cargo test -p skippy-server --lib (447 passed, 3 ignored)

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: d2c2a079-6604-40e1-aa99-adf75f28244e

📥 Commits

Reviewing files that changed from the base of the PR and between 0d12af5 and a0dfc85.

📒 Files selected for processing (2)

• crates/skippy-server/src/frontend/request.rs
• crates/skippy-server/src/frontend/tests/request.rs

Included review availability: Your plan includes up to 8 reviews per rolling hour; 6 remain after this review.

---

📝 Walkthrough

Walkthrough

The request parser now defaults unspecified chat reasoning formats to Auto instead of Hidden. Request tests update canonical, reasoning-effort, provider, and kwargs expectations and verify that the default template exposes reasoning.

Changes

Chat reasoning format

Layer / File(s)	Summary
Default reasoning format and validation crates/skippy-server/src/frontend/request.rs, crates/skippy-server/src/frontend/tests/request.rs	The parser defaults an unspecified reasoning format to Auto. Tests cover template defaults and reasoning-effort, provider, and kwargs overrides.

Estimated code review effort: 1 (Trivial) | ~5 minutes

Merge Risk: ⚪ Minimal · up to a0dfc

The change updates the default reasoning-output behavior while preserving explicitly configured formats, with no actionable merge-blocking risk remaining after normal checks and review.

Possibly related PRs

• Mesh-LLM/mesh-llm#947: Modifies the same reasoning-format mapping and changes the unset default from Hidden to Auto.
• Mesh-LLM/mesh-llm#1153: Updates reasoning-default handling and related request tests.
• Mesh-LLM/mesh-llm#1211: Changes the same chat template reasoning behavior and related tests.

Suggested reviewers: i386

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
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.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly describes the primary user-facing change: preventing blank responses from thinking models.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch jimmy/fix-reasoning-format-auto-default

---

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.}

[ndizazzo]

big oof, as the kids say