validtr run

Run task analysis, recommendation, execution, testing, scoring, and retries.

Usage

validtr run "Build a FastAPI app with JWT auth"

Flags

• --provider: anthropic, openai, gemini
• --compare: comma-separated providers
• --dry-run: recommendation only
• --model: override model id
• --max-retries: retry limit
• --score-threshold: pass threshold
• --timeout: execution timeout seconds

Modes

Single provider

validtr run "Build a CLI in Go" --provider openai

Compare

validtr run "Build CRUD API" --compare anthropic,openai,gemini

Behavior:

• providers run sequentially
• provider with missing API key is skipped

Dry-run

validtr run "Automate PR code reviews" --dry-run

Behavior:

• no execution or test containers
• returns analyzed task + recommended stack

Environment

• provider API key must be set for selected provider
• optional TAVILY_API_KEY enriches recommendation search