fix: respect user's model selection instead of forcing overrides#1050
Open
srobinson wants to merge 1 commit intoruvnet:mainfrom
Open
fix: respect user's model selection instead of forcing overrides#1050srobinson wants to merge 1 commit intoruvnet:mainfrom
srobinson wants to merge 1 commit intoruvnet:mainfrom
Conversation
Fixes ruvnet#1044 — user sets Sonnet 4.5 in Claude Code but claude-flow forces Opus 4.5. Three root causes: 1. Wrong Opus model ID: `claude-opus-4-20250514` doesn't exist, corrected to `claude-opus-4-5-20251101` 2. ANTHROPIC_MODEL env var unconditionally overwritten in executeClaudeCode(), ignoring any user-set value. Now only sets the env var when the user hasn't already configured one. 3. headless.ts runtime hardcoded `model: 'sonnet'` as a config override, stomping each worker's own model preference. Removed so workers use their configured model (or fall back to sonnet). Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
There was a problem hiding this comment.
Pull request overview
This PR fixes issue #1044 where users configure Sonnet 4.5 in Claude Code but claude-flow incorrectly forces Opus 4.5. The fix addresses three root causes: correcting an invalid Opus model ID, respecting user-set ANTHROPIC_MODEL environment variables, and removing hardcoded model overrides.
Changes:
- Fixed Opus model ID from non-existent
claude-opus-4-20250514toclaude-opus-4-5-20251101 - Modified environment variable handling to respect existing ANTHROPIC_MODEL values set by users
- Removed hardcoded
model: 'sonnet'override that was preventing workers from using their configured models
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
| v3/@claude-flow/cli/src/services/headless-worker-executor.ts | Corrected Opus 4.5 model ID and added conditional logic to preserve user's ANTHROPIC_MODEL env var |
| v3/@claude-flow/cli/src/runtime/headless.ts | Removed hardcoded model parameter to allow workers to use their configured default models |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+1122
to
+1124
| if (!env.ANTHROPIC_MODEL) { | ||
| env.ANTHROPIC_MODEL = MODEL_IDS[options.model]; | ||
| } |
There was a problem hiding this comment.
The logic for respecting the user's ANTHROPIC_MODEL environment variable is not covered by tests. Consider adding a test case that verifies:
- When ANTHROPIC_MODEL is already set in process.env, it should not be overridden
- When ANTHROPIC_MODEL is not set, it should default to the worker's configured model
This would help prevent regressions and document the intended behavior.
srobinson
added a commit
to srobinson/fmm
that referenced
this pull request
Jan 30, 2026
Live demo: used fmm to fix a bug in ruvnet/claude-flow (9,008 files). fmm init indexed 2,221 files in 3s, navigated straight to 2 files, 5-line fix, clean PR pushed without fmm artifacts. - Case study with diffs, metrics, workflow documentation - Social content: 5-tweet thread + short-form blog post PR: ruvnet/ruflo#1050 Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2 tasks
srobinson
added a commit
to srobinson/fmm
that referenced
this pull request
Jan 30, 2026
* exp16: A/B cost experiment + fix dep_matches and .ts/.js priority Bug fixes: - dep_matches now resolves relative paths (../foo) against dependent file's directory instead of naive suffix matching. Fixes downstream dependents always returning empty. - Export index prefers .ts over .js when both export the same symbol, preventing non-deterministic race in parallel manifest generation. Experiment (exp16-cost): - Docker-isolated A/B: vanilla Claude vs Claude+fmm - 10 verifiable tasks with ground truth from fmm index - Result: fmm eliminates grep (-100%), fewer tools (-33%), but only -10% tokens — system prompt overhead drowns real navigation cost. - Proves Docker API approach insufficient; need CLI-based test. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * Pivot to .fmm sidecar files, drop inline frontmatter Replace inline `// --- FMM ---` comment blocks with `.fmm` sidecar companion files (e.g., `foo.rs` → `foo.rs.fmm`). Sidecars use plain YAML, are independent of source file format, and don't pollute source. - Extractor writes sidecar files instead of prepending to source - CLI: remove `--manifest-only` flag, add `clean` subcommand - Formatter outputs YAML sidecar format instead of comment blocks - Manifest reads from sidecars, drops inline parsing - MCP serves sidecar-based data - Navigation skill rewritten for sidecar-first workflow - Fixtures and tests updated for new format - Remove obsolete docs (CLAUDE-SNIPPET.md, claude-skill.md) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * Add claude-flow #1044 case study — real-world fmm proof Live demo: used fmm to fix a bug in ruvnet/claude-flow (9,008 files). fmm init indexed 2,221 files in 3s, navigated straight to 2 files, 5-line fix, clean PR pushed without fmm artifacts. - Case study with diffs, metrics, workflow documentation - Social content: 5-tweet thread + short-form blog post PR: ruvnet/ruflo#1050 Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * Add comprehensive research docs + fix score.py review issues 6 research docs: architecture, CLI reference, data formats, MCP integration, experiment results, skill integration. Fix Copilot review: remove unused sys import, close file handles. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Fixes #1044 — user configures Sonnet 4.5 in Claude Code but claude-flow forces Opus 4.5.
Three root causes found and fixed:
claude-opus-4-20250514→claude-opus-4-5-20251101(the old ID doesn't exist)ANTHROPIC_MODELwas unconditionally overwritten inexecuteClaudeCode(), ignoring any user-set value. Now respects existing env var.headless.tsruntime passedmodel: 'sonnet'as a config override, stomping each worker's own model preference. Removed so workers use their configured model.Changes
v3/.../headless-worker-executor.tsANTHROPIC_MODELenv varv3/.../headless.tsmodel: 'sonnet'overrideTest plan
ANTHROPIC_MODEL=claude-sonnet-4-20250514in env, run a headless worker — should use Sonnet, not Opusultralearnanddeepdiveworkers should correctly resolve to Opus 4.5 (new model ID)🤖 Generated with Claude Code