The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill documents integration with the Codecov coverage reporting service using its official bash uploader.
Evidence: The GitHub Actions example in SKILL.md uses the command bash <(curl -s https://codecov.io/bash) to upload coverage data.
Context: Codecov is a well-known developer service, and this pattern is a standard (though documented) method for CI/CD integration.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied C++ source code to generate unit tests, which creates a surface for indirect instructions to be processed by the agent.
Ingestion points: The code_to_test property in the input_schema defined in SKILL.md accepts arbitrary strings of C++ code.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are specified in the prompt schema.
Capability inventory: The skill is primarily instructional and generates code templates for testing, benchmarking, and fuzzing; it does not perform automated execution of the generated code.
Sanitization: The skill does not define specific validation or sanitization rules for the input code contents.

testing