The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external code (e.g., AI-generated snippets or git diffs), which introduces a potential attack surface for indirect prompt injection. Malicious instructions could be embedded within code comments or string literals in the source files being analyzed to influence the agent's behavior during the 'degunking' process.
Ingestion points: The skill processes user-provided code, file modules, and branch diffs via 'Full Review Mode' and 'Branch Review Mode'.
Boundary markers: The instructions do not specify the use of strict delimiters or boundary markers to separate instructional content from the data being analyzed.
Capability inventory: The agent is authorized to rewrite code, simplify logic, and execute version control commands such as git diff to perform its task.
Sanitization: No explicit sanitization or filtering logic is provided to handle instructions embedded within the target code.

code-degunker