ml-research
Native Claude Code plugin for research-first autonomous ML engineering and fine-tuning (SFT/DPO/GRPO/LoRA) on Hugging Face Jobs.
🛡️ AgentReady threat assessment
MAESTRO 7-layer threat model + OWASP AIVSS risk score for ml-research, derived from its capabilities.
These scores are auto-generated from public information (the agent's own listing, docs, and repository) using the canonical OWASP AIVSS formula and the MAESTRO framework — an estimate for guidance, not a penetration test, audit, or certification. See the scoring methodology — every score is re-derived by the same automated method as an agent's public evidence changes.
Overview
A Claude Code port of Hugging Face's ml-intern that turns Claude into an autonomous ML engineer. It bundles 2 skills, 2 agents, and 1 hook for dataset/repo/paper tooling and runs an optional budget-bounded autonomous experiment sweep. Every model call runs on the user's Claude Code subscription (no separate LLM API key), executing fine-tuning jobs remotely on HF Jobs.
Key features and capabilities
- Research-first fine-tuning (SFT/DPO/GRPO/LoRA) on Hugging Face Jobs
- Optional budget-bounded autonomous experiment loop
- No LLM API key required — runs on Claude Code subscription
Use cases
- Autonomous fine-tuning of open models on HF infrastructure
- Running budget-capped ML research experiments from Claude Code