Skip to content

feat: disable reranking in bundled backend - #151

Merged
yyiilluu merged 1 commit into
mainfrom
codex/service-only-inference
Aug 7, 2026
Merged

feat: disable reranking in bundled backend#151
yyiilluu merged 1 commit into
mainfrom
codex/service-only-inference

Conversation

@yyiilluu

@yyiilluu yyiilluu commented Aug 7, 2026 •
edited by coderabbitai Bot
Loading

Copy link
Copy Markdown
Contributor

Summary

  • Disable Reflexio's reranker feature in claude-smart's bundled OSS backend so the colocated inference service does not load or prewarm an unused cross-encoder.
  • Clarify that the dashboard reranker toggle applies only when connected to a Reflexio deployment with reranking enabled.

Changes

  • Hard-sets REFLEXIO_RERANK_ENABLED=false in backend-service.sh, after loading the managed environment, so shared environment values cannot accidentally enable model residency.
  • Updates the dashboard copy and developer documentation for the shared-service inference architecture.
  • Adds a regression assertion covering the bundled-backend feature flag.

Test Plan

  • pytest tests/test_codex_support.py: 39 passed
  • Dashboard ESLint: passed
  • Dashboard TypeScript type check: passed

Summary by CodeRabbit

  • New Features

    • Added a configuration option to control search reranking for connected deployments.
    • Clarified that reranking uses a shared cross-encoder relevance service when enabled.

  • Bug Fixes

    • The bundled backend now consistently disables cross-encoder reranking, preventing unnecessary model loading and startup prewarming.

  • Documentation

    • Updated configuration guidance to reflect the new reranking setting and behavior.

@yyiilluu
feat: disable reranking in bundled backend
ba3c02f 
Keep claude-smart's shared inference service lightweight by hard-disabling reranker loading while documenting that the dashboard toggle only applies to connected deployments.
@coderabbitai

coderabbitai Bot commented Aug 7, 2026 •
edited
Loading

Copy link
Copy Markdown

Review Change Stack

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info
⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 05c63856-56f5-4610-b06a-a088f63894ef

📥 Commits

Reviewing files that changed from the base of the PR and between d096b3b and ba3c02f.

📒 Files selected for processing (4)
  • DEVELOPER.md
  • plugin/dashboard/app/configure/server/page.tsx
  • plugin/scripts/backend-service.sh
  • tests/test_codex_support.py
📝 Walkthrough

Walkthrough

The bundled backend now exports REFLEXIO_RERANK_ENABLED=false. Developer documentation, server configuration text, and a Codex hook test reflect the disabled in-process reranker behavior.

Changes

Reranker disablement

Layer / File(s) Summary
Backend reranker enforcement
plugin/scripts/backend-service.sh, tests/test_codex_support.py		 The backend exports REFLEXIO_RERANK_ENABLED=false. The Codex hook test verifies the export.
Reranker configuration documentation
DEVELOPER.md, plugin/dashboard/app/configure/server/page.tsx		 Documentation describes shared-service cross-encoder behavior and the disabled claude-smart backend model.

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

Possibly related PRs

Suggested reviewers: yilu331, wenchanghan

🚥 Pre-merge checks | ✅ 5
✅ Passed checks (5 passed)
Check name Status Explanation
Description Check ✅ Passed Check skipped - CodeRabbit’s high-level summary is enabled.
Title check ✅ Passed The title clearly and concisely describes the main change: disabling reranking in the bundled backend.
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.
✨ 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 codex/service-only-inference

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

@yyiilluu
yyiilluu merged commit 08335bb into main Aug 7, 2026
9 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@yyiilluu