The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to execute arbitrary shell commands defined in the workspace's .evolve/config.yaml file for the agent and evaluator components. Evidence: Configuration examples show usage such as run: python3 ./evaluate.py and command: node.
[EXTERNAL_DOWNLOADS]: The gepa strategy utilizes the uv package manager to fetch and execute specific versions of Python packages, including gepa==0.1.1 and litellm==1.82.6, at runtime.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface due to its core functionality of ingesting and analyzing workspace data to generate new code proposals. 1. Ingestion points: Workspace files (e.g., packing.py) and historical records stored in the .evolve/runtime/ directory. 2. Boundary markers: The documentation does not specify the use of strict boundary markers or instructions to ignore instructions embedded within the processed data. 3. Capability inventory: The skill can execute shell commands via the evolve run and evolve validate commands and performs file system modifications. 4. Sanitization: No explicit sanitization or validation of the ingested content is documented.

evolve-cli