$ show NVIDIA-BioNeMo/bionemo-agent-toolkit
This repository is the NVIDIA BioNeMo Agent Toolkit — a collection of 31 skills that give coding/scientific agents (Claude Code, Codex, any SKILL.md-compatible harness) domain knowledge for life-science workflows: protein structure prediction, molecular docking, generative chemistry, and related tasks. Skills execute in the user's own environment (workstation, notebook, cluster) and wrap NVIDIA NIM (NVIDIA Inference Microservices) APIs — both hosted endpoints and local Docker deployments. The toolkit is not an installable Python package; it is a skill catalog with supporting tooling, dual-licensed (Apache-2.0 OR CC-BY-4.0).
The LLM identified 3 files worth highlighting for reuse:
Self-contained Python script enforcing synchronization between source skills and the generated aggregate `skills/bionemo-agent-toolkit/`. Solves the recurring problem of generated-artifact drift in monorepos. Reusable in any project that generates aggregate artifacts from source-of-truth components.
Bash preflight script validating GPU availability, environment variables (AF2_DIR, RF3_CKPT_PATH, etc.), and required tool paths before running GPU-backed skills. A clean, self-contained environment-gating pattern reusable for any ML/GPU skill or pipeline that needs to fail fast with actionable diagnostics.
The common NIM skill pattern across all skills: SKILL.md for mode selection (hosted/local), auth via NGC_API_KEY/NVIDIA_API_KEY with .env fallback, local Docker with LOCAL_NIM_CACHE, readiness polling, Python requests POST with Bearer auth, and artifact saving. This pattern is the core reusable asset — it can be extracted as a template for any NIM-backed agent skill.