The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several shell commands to support the TDD workflow.
Runs rg (ripgrep) to discover existing test patterns and files within the repository.
Executes cargo test and cargo tarpaulin to verify implementation and measure code coverage.
Uses ralph emit (a vendor-specific tool) to report build and test status.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core functionality.
Ingestion points: Reads and parses content from .spec.md and .code-task.md files to extract acceptance criteria.
Boundary markers: The instructions do not specify any delimiters or safety warnings to ignore instructions embedded within the specification files.
Capability inventory: The skill has the ability to generate code stubs based on external input and execute them via cargo test, as well as run search and reporting commands.
Sanitization: There is no mention of sanitizing or validating the content of the specification files before using them to generate executable code. An attacker could embed malicious instructions in a specification file that the agent might inadvertently follow during the code generation or test execution phases.

test-driven-development