The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill analyzes untrusted source code to generate reviews and test cases.
Ingestion points: SKILL.md describes reading code via git diff or user-provided file paths.
Boundary markers: Absent. The skill does not define specific delimiters for the untrusted code content.
Capability inventory: The skill has the ability to read files, write new test files to the project directory, and execute shell commands to run those tests.
Sanitization: No sanitization of the processed code is mentioned before it is used to influence the agent's output and test generation.
[DYNAMIC_EXECUTION]: The skill generates and runs test scripts based on the results of the code review.
Evidence: Instructions to write qc-* test files to the project and execute python3 harness/scripts/qc-regression.py --run.
[COMMAND_EXECUTION]: The skill relies on local shell commands for version control analysis and test execution.
Evidence: Use of git diff for scope determination and python3 for running regression and grounding scripts.

qc-code