The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several local commands and scripts, including ./scripts/format.sh, ./scripts/validate.sh, curl, jq, docker logs, and git diff. These operations are used for repository maintenance, health checks, and formatting.
[EXTERNAL_DOWNLOADS]: The skill facilitates running evaluations via python -m evals, which connects to OpenAI. This action is explicitly gated by a user confirmation prompt and interacts with a well-known service provider for standard LLM benchmarking.
[PROMPT_INJECTION]: The skill processes untrusted repository content (ingestion points include all documentation and code), which creates a surface for indirect prompt injection. It has the capability to write to files and execute shell commands (inventory includes file-write and subprocess calls). No specific boundary markers or sanitization logic are defined to prevent the agent from potentially obeying instructions embedded in the analyzed files.

review-and-improve