Skip to content

fix: remove "minimal" thinking level (LLM-2595) - #914

Closed
cast-vytautas wants to merge 3 commits into
getkimchi:masterfrom
cast-vytautas:fix/remove-minimal-thinking-level
Closed

fix: remove "minimal" thinking level (LLM-2595)#914
cast-vytautas wants to merge 3 commits into
getkimchi:masterfrom
cast-vytautas:fix/remove-minimal-thinking-level

Conversation

@cast-vytautas

@cast-vytautas cast-vytautas commented Jul 22, 2026

Copy link
Copy Markdown
Contributor

Linked issue

Closes #915

What does this PR do?

Removes the minimal thinking level from the harness. This level causes errors with vLLM/sglang providers (LLM-2595). The proxy already maps minimal to low; this stops the harness from offering a broken level.

Changes:

  • Remove minimal from ThinkingLevel type union (personas/types.ts)
  • Remove minimal from THINKING_LEVEL_ORDER array; change DELEGATION_THINKING_BASE.explore.simple from minimal to low; update renderDelegationThinkingLevelTable() Explore row (thinking-level-policy.ts)
  • Remove minimal from Agent tool schema description and agent config help text (agents/index.ts)
  • Remove minimal from agent wizard thinking-level select menu (agents/index.ts)
  • Remove minimal from set_phase tool thinking parameter enum (tags.ts)
  • Remove minimal from CLI --thinking help text (commands/help.ts)
  • Update THINKING_LEVELS prose in orchestration instructions (orchestration-instructions.ts)
  • Update docs/agents.md frontmatter reference
  • Update invocation-config.test.ts to use low instead of minimal
  • Update thinking-level-policy.test.ts assertions and add regression test for absence of minimal
  • Add assertion in orchestration-instructions.test.ts that rendered instructions do not contain minimal

Checklist

  • I have read CONTRIBUTING.md and agree to the CLA
  • This PR links to an open issue above
  • Tests pass locally (pnpm run test)
  • Lint passes (pnpm run check)
  • Documentation updated if behavior changed

@kimchi-review

kimchi-review Bot commented Jul 22, 2026

Copy link
Copy Markdown

Kimchi Code Review

Property Value
Commit 0ff890b
Author @cast-vytautas
Files changed 0
Review status Completed
Comments 3 (1 info, 2 warning)
Duration 42s

Summary

📊 Review Score: 80/100 (overall code quality — 0 lowest, 100 highest)
⏱️ Estimated effort to review: 1/5 (1 = trivial, 5 = very complex)

🧪 Tests: yes — Existing tests were updated to replace minimal with low, and a new test was added to verify orchestration instructions no longer mention minimal. Coverage is appropriate for the scope but does not verify runtime rejection of the removed value.

📝 Found 3 issue(s). See inline comments for details.

What to expect

Kimchi will analyze the changes in this pull request and post:

  • A summary of the overall changes
  • Inline comments on specific lines with findings categorized by issue type

The review typically completes within a few minutes. This comment will be updated once the review is ready.

Interact with Kimchi
  • @getkimchi review — re-trigger a full review on the latest commit
  • @getkimchi summary — regenerate the PR summary
  • @getkimchi ignore — skip this PR (no review will be posted)
  • Reply to any inline comment to ask follow-up questions or request clarification
Configuration

Reviews are configured by your organization admin.
Review instructions, excluded directories, and severity thresholds can be adjusted per repository in the Kimchi dashboard.


Powered by Kimchi — AI-powered code review by CAST AI

@readme-ai-writer

readme-ai-writer Bot commented Jul 22, 2026

Copy link
Copy Markdown

📚 No documentation changes were needed for this PR.

@kimchi-review kimchi-review Bot left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

📊 Review Score: 80/100 (overall code quality — 0 lowest, 100 highest)
⏱️ Estimated effort to review: 1/5 (1 = trivial, 5 = very complex)

🧪 Tests: yes — Existing tests were updated to replace minimal with low, and a new test was added to verify orchestration instructions no longer mention minimal. Coverage is appropriate for the scope but does not verify runtime rejection of the removed value.

📝 Found 3 issue(s). See inline comments for details.

Comment thread src/extensions/agents/index.ts
Comment thread src/extensions/agents/personas/types.ts
Comment thread src/extensions/orchestration/orchestration-instructions.test.ts Outdated
The "minimal" thinking level causes errors with vLLM/sglang providers.
The proxy already maps minimal → low; this stops the harness from
offering a broken level.

Changes:
- Remove "minimal" from ThinkingLevel type union
- Remove "minimal" from agent tool schema descriptions
- Remove "minimal" from agent wizard thinking-level select
- Remove "minimal" from CLI --thinking help text
- Update docs/agents.md frontmatter reference
- Update invocation-config test to use "low" instead of "minimal"
- Add assertion that orchestration instructions don't contain "minimal"
@cast-vytautas
cast-vytautas force-pushed the fix/remove-minimal-thinking-level branch from 0ff890b to 870a713 Compare July 22, 2026 13:10
cast-vytautas and others added 2 commits July 22, 2026 16:21
Address kimchi-review bot comments on PR getkimchi#914:

1. Replace Type.String() with Type.Union of Type.Literal for the
   Agent tool's thinking parameter — runtime enum validation now
   rejects invalid values.

2. Add coerceThinkingLevel() in custom-agents.ts that maps the removed
   'minimal' level to 'low' for backward compatibility with existing
   agent profile YAML files. Applied in both custom-agents.ts (YAML
   loading) and invocation-config.ts (runtime param resolution).

3. Strengthen the orchestration-instructions test to also assert that
   the valid level string 'off, low, medium, high, xhigh' is present
   in the rendered output, not just that 'minimal' is absent.

Co-Authored-By: Kimchi <noreply@kimchi.dev>
@github-actions

github-actions Bot commented Aug 6, 2026

Copy link
Copy Markdown

This PR has been marked as stale because it has not had recent activity. It will be closed tomorrow if no further activity occurs.

@github-actions github-actions Bot added the stale No recent activity label Aug 6, 2026
@github-actions github-actions Bot closed this Aug 8, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

stale No recent activity

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Remove "minimal" thinking level that causes errors with vLLM/sglang providers

1 participant