The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Downloads the mergekit toolkit from its official GitHub repository and fetches machine learning libraries from standard package registries.
[COMMAND_EXECUTION]: Instructs the agent to execute shell commands for repository cloning, environment configuration, and model merging operations via CLI tools.
[DYNAMIC_EXECUTION]: Programmatically generates configuration files at runtime which are subsequently used as parameters for external subprocess execution during the merging process.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied evaluation prompts to generate model responses for consistency-based coefficient selection.
Ingestion points: Evaluation prompt arrays consumed by inference functions in the coefficient tuning reference file.
Boundary markers: No specific delimiters or boundary markers are used when interpolating external prompts into model context.
Capability inventory: Includes model inference capabilities and shell execution access for the merge toolkit.
Sanitization: No input validation or content filtering is performed on the evaluation prompts.

model-merging