The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the project's source code, git diffs, and git logs which could contain malicious instructions designed to influence the agent's behavior during the analysis.
Ingestion points: Project source files, git diff output, git status, and git log output are read to identify undertested logic and past regressions (SKILL.md Steps 1, 2, and 4).
Boundary markers: The instructions do not define delimiters or specific warnings to ignore instructions embedded within the analyzed code or commit messages.
Capability inventory: The agent has access to Bash, Edit, Write, and Task tools to execute tests and modify the filesystem (SKILL.md frontmatter).
Sanitization: There is no evidence of sanitization, filtering, or validation of the ingested external data before it is processed by the agent.
[COMMAND_EXECUTION]: The skill is designed to execute the project's native test suite, which involves running arbitrary commands and code defined within the project environment.
Evidence: The skill explicitly instructs the agent to "Detect the test framework and runner" and "Run the project's real test command" (SKILL.md Steps 1 and 8).
[DYNAMIC_EXECUTION]: The skill facilitates the execution of tests that may be dynamically generated or modified by the agent itself based on its analysis of the source code.
Evidence: Step 7 in SKILL.md describes the process of writing new tests to the working tree and running them to verify behavior.

test-coverage