The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes user-specified shell commands provided via the --test-command argument. It also spawns a headless AI agent instance (claude -p) with access to the Bash tool to perform file edits and verification.
[PROMPT_INJECTION]: The skill processes the content of local files specified in the --test argument, which introduces a potential surface for indirect prompt injection where adversarial test content could influence the AI agent's actions.
Ingestion points: Content of the test file path provided to the --test parameter (SKILL.md).
Boundary markers: No explicit delimiters or instructions are documented to isolate test data from the agent's primary instructions.
Capability inventory: The spawned agent has access to Read, Edit, and Bash tools (SKILL.md).
Sanitization: The skill does not describe any sanitization or validation of the ingested file content.
[EXTERNAL_DOWNLOADS]: The documentation references the ruvnet/agent-harness-generator repository on GitHub as a design reference.

tdd-repair