apprentice [--version] [--config PATH] <command>
--version— print version and exit--config PATH— path toapprentice.toml(default:config/apprentice.toml)
Run the pipeline through review (no packaging).
apprentice build <algorithm> [--tier N] [--description TEXT] [--backend NAME] [--model STRING]
algorithm— algorithm name (e.g. "quicksort")--tier— algorithm tier 1-4 (default: 2)--description— optional description for the LLM--backend— override provider backend (anthropic, openai, gemini, ollama, local)--model— override LiteLLM model string (e.g. "ollama_chat/llama3.3")
Persists run state to ~/.apprentice/sessions/ for retry support.
Run the full pipeline including packaging (PR creation).
apprentice submit <algorithm> [--tier N] [--backend NAME] [--model STRING]
Same options as build. Creates PRs in both no-magic and no-magic-viz repos.
Run the discovery agent to suggest candidate algorithms.
apprentice suggest [--tier N] [--limit N] [--backend NAME] [--model STRING]
--tier— target tier (default: 2)--limit— max candidates to suggest (default: 5)
Retry a failed pipeline run.
apprentice retry <run_id> [--backend NAME] [--model STRING]
run_id— ID fromapprentice historyoutput
Reruns the full pipeline for the same algorithm and tier.
List past pipeline runs.
apprentice history [--status STATUS] [--limit N]
--status— filter by status: completed, failed, in_progress--limit— max entries (default: 20)
Show aggregated metrics across all recorded runs.
apprentice metrics
Reports success rate, per-agent cost/token breakdown, and per-tier statistics.
Inspect artifacts from the last build.
apprentice preview
Show budget usage and system state.
apprentice status
Display current configuration.
apprentice config
Launch ADK dev UI for interactive debugging.
apprentice dev [--port N]
--port— dev UI port (default: 8080)