The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill correctly uses the hf_jobs MCP tool to execute training scripts on Hugging Face's ephemeral infrastructure. This is the intended purpose of the skill and follows secure implementation patterns.
[EXTERNAL_DOWNLOADS]: Fetches model conversion tools and training templates from trusted sources, including Hugging Face's official GitHub repositories and the well-known llama.cpp repository.
[COMMAND_EXECUTION]: Utilizes subprocess.run to install build dependencies (such as cmake and build-essential) and run model quantization binaries within the isolated job environment. This is required for its primary function of GGUF conversion.
[INDIRECT_PROMPT_INJECTION]: The skill provides an interface to ingest and process external datasets from the Hugging Face Hub. Ingestion points: datasets.load_dataset calls in training scripts. Boundary markers: None. Capability inventory: Execution of Python training code via the hf_jobs tool. Sanitization: Relies on the standard Hugging Face datasets library processing. This represents a standard surface for a model-training tool.

model-trainer