fix(claude-code): pass configured model to the spawned CLI (#2881) - #7
Closed
chethanuk wants to merge 2 commits into
Closed
fix(claude-code): pass configured model to the spawned CLI (#2881)#7chethanuk wants to merge 2 commits into
chethanuk wants to merge 2 commits into
Conversation
…ectorize-io#2881) Red test: both call() and call_with_tools() must place self.model on the ClaudeAgentOptions handed to the Claude Agent SDK, since that field is what the SDK transport turns into the CLI --model flag. Currently the model is dropped, so the spawned claude CLI runs its own default while metrics report self.model.
…-io#2881) ClaudeCodeLLM.call() and call_with_tools() built ClaudeAgentOptions without model=, so the model resolved by config (global, per-scope, or the provider default) was stored on self.model and reported in metrics/traces but never reached the spawned `claude` CLI — it silently ran its own default model. ClaudeAgentOptions.model is the only field the SDK transport turns into the CLI's --model flag; add model=self.model at both option sites (unconditional, the transport already skips a falsy model). Move the claude-code provider default from the dated claude-sonnet-4-5-20250929 to the undated alias claude-sonnet-4-5 (one of the ids the Claude Agent SDK documents as a model example, and the "full name" form the CLI accepts). The dated id was previously inert; now that it becomes a real --model argument, an undated form avoids breaking provider-only setups such as the shipped docker-compose claude-code profile. Aligned across config.py, llmProviders.json, and the configuration docs; skills/ references regenerated. Add the model field to the _FakeOptions stand-ins in the two existing claude-code test modules so they keep constructing options once model= is passed.
|
Caution The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased. |
🤖 CodeAnt AI — Review Status
|
Thanks for using CodeAnt! 🎉We're free for open-source projects. if you're enjoying it, help us grow by sharing. Share on X · |
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (9)
📝 WalkthroughWalkthroughThe Claude Code provider now passes its configured model explicitly to the Claude Agent SDK for regular and tool-enabled calls. Default model references and documentation use ChangesClaude Code model selection
Estimated code review effort: 2 (Simple) | ~10 minutes Suggested labels: Suggested reviewers: 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches📝 Generate docstrings
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. Comment |
![codeant-ai[bot]](https://avatars.githubusercontent.com/in/646884?s=60&v=4){kind=small}
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
1 participant
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.
Fixes the behaviour reported in vectorize-io#2881.
Problem
With
HINDSIGHT_API_*_LLM_PROVIDER=claude-code, the model resolved from config — global (HINDSIGHT_API_LLM_MODEL), per-scope (..._RETAIN_/_REFLECT_/_CONSOLIDATION_LLM_MODEL), or the provider default — was accepted, stored on the provider asself.model, and reported in metrics/traces, but the spawnedclaudeCLI ignored it and ran its own default model. Cost/latency attribution was wrong and per-scope model tuning silently did nothing.Root cause
ClaudeCodeLLM.call()andClaudeCodeLLM.call_with_tools()inhindsight-api-slim/hindsight_api/engine/providers/claude_code_llm.pybuiltClaudeAgentOptions(...)withoutmodel=.ClaudeAgentOptions.modelis the only field the Claude Agent SDK transport turns into the CLI's--modelflag, so the configured model never reached the subprocess.Fix
model=self.modelat bothClaudeAgentOptionssites (unconditional — the SDK transport already skips a falsy model).claude-codeprovider default from the datedclaude-sonnet-4-5-20250929to the undated aliasclaude-sonnet-4-5(an id the Claude Agent SDK documents as amodelexample and the "full name" form the CLI accepts). The dated id was previously inert; now that it becomes a real--modelargument, the undated form avoids breaking provider-only setups such as the shipped docker-composeclaude-codeprofile. Aligned acrossconfig.py,hindsight-docs/src/data/llmProviders.json, and the configuration docs;skills/references regenerated.modelfield to the_FakeOptionsstand-ins in the two existing claude-code test modules so they keep constructing options oncemodel=is passed.Test evidence
New parametrized regression test
tests/test_claude_code_llm_model.pymocks the SDK and asserts the configured model lands onClaudeAgentOptionsfor bothcallandcall_with_tools(rows:claude-haiku-4-5,sonnet,claude-sonnet-4-5). It fails 6/6 before the fix (AssertionError: assert None == '<model>') and passes after.ruff check .— All checks passedruff format --check(touched files) — already formattedty check(provider + config) — All checks passed./scripts/generate-docs-skill.sh— regeneratedskills/in sync (no diff)Note:
./scripts/hooks/lint.shESLint step could not run in the QA environment (@eslint/jsnot installed — control-planenode_modulesabsent). This branch changes zero JS/TS files, so ESLint has nothing of this change to check.Refs vectorize-io#2881