The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various shell commands including git, grep, ls, and mkdir to analyze the repository context, detect the tech stack, and generate documentation.
[COMMAND_EXECUTION]: The skill facilitates the building and execution of local code from the repository being analyzed (e.g., cargo build, npm run build, ./target/debug/...). This is expected behavior for an E2E testing utility.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface as it ingests untrusted data from the repository (branch names, commit messages, and file contents) and interpolates it into a YAML test plan.
Ingestion points: Data enters the context via git rev-parse, git diff, git log, and grep commands in SKILL.md (Steps 1, 3, and 4).
Boundary markers: None identified; data is interpolated directly into docs/testing/test-plan.yaml fields like changes_summary and context.
Capability inventory: The skill can execute shell commands, write files to the disk, and trigger project builds using standard package managers.
Sanitization: No explicit sanitization or escaping of the repository metadata is performed before interpolation into the YAML plan.
[EXTERNAL_DOWNLOADS]: The skill triggers dependency downloads through official package registries (npm, PyPI, Cargo, Mix, Go) when executing project setup commands. These are well-known services and represent standard development workflows.

gen-test-plan