What problem does it solve?
Onboard and profile a CLI tool to create reusable environment profiles for AI agent evaluations, ensuring consistent runtime, OS constraints, and non-interactive flag discovery for reuse by evaluation skills.
Core Features & Use Cases
- Read agent-facing docs (AGENTS.md, CODING_AGENTS.md, README.md) to determine canonical invocation, env vars, and non-interactive flags.
- Detect runtime and toolchain from common manifests (pyproject.toml/setup.py for Python, package.json for Node, Cargo.toml for Rust, go.mod for Go) and map to a preferred runner.
- Locate and validate the target binary by attempting <cli-name> --version and --help, then resolve path for artifact generation.
- Discover non-interactive flags and config options to enable deterministic evaluations.
- Save a local environment artifact named <cli-name>-environment containing OS, runtime, binary, version, non-interactive flags, and relevant config env vars.
Quick Start
Provide the CLI name or path, run the onboarding skill against it, and the tool will produce a reusable <cli-name>-environment artifact for subsequent evaluations.