The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess.run to execute testing frameworks (e.g., pytest, jest, vitest, go test, cargo test) and git commands. This functionality is the primary purpose of the skill and is required for TDD automation. The commands are constructed as argv lists without using shell=True, which follows security best practices.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and parses stdout/stderr from external test processes to classify test outcomes (e.g., identifying semantic failures vs. syntax errors).
Ingestion points: Untrusted data enters the agent context via scripts/test_runner.py, which captures the full output of subcommands.
Boundary markers: None are present. The test output is returned as a raw string within the artifacts field of the JSON output.
Capability inventory: The skill has capabilities to execute shell commands (subprocess.run) and write to the local filesystem for state persistence (scripts/tdd_state_machine.py).
Sanitization: The skill applies basic truncation to the captured test output (typically limiting it to the first 2000 characters) before returning it to the agent, but it does not perform instruction filtering or sanitization.

sc-tdd