Conversation
Add model: haiku to the Task spec in the e2e test to reduce costs by using the cheapest available model instead of the default.
There was a problem hiding this comment.
Pull request overview
This PR modifies the e2e test to explicitly specify a cheaper Claude model ("haiku") instead of using the default model, with the goal of reducing test execution costs.
Changes:
- Added
model: haikufield to the Task specification in the e2e test
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| name: ` + taskName + ` | ||
| spec: | ||
| type: claude-code | ||
| model: haiku |
There was a problem hiding this comment.
The model name "haiku" may not match the expected format for Claude Code models. Based on the README.md documentation (line 106) and integration test (test/integration/task_test.go:65), model identifiers should use the full format like "claude-sonnet-4-20250514" or similar versioned names. Using just "haiku" might not be recognized by the Claude Code CLI. Consider using a complete model identifier like "claude-3-5-haiku-20241022" or verifying that "haiku" is a valid short form accepted by the CLI.
| model: haiku | |
| model: claude-3-5-haiku-20241022 |
Summary
model: haikuto the Task spec in the e2e test to use the cheapest available model instead of the default, reducing test costsTest plan
make testpassesmake test-e2eruns with haiku model (requires cluster +CLAUDE_CODE_OAUTH_TOKEN)🤖 Generated with Claude Code