The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze content from the user's codebase (TypeScript and test files) to identify and execute code deletions and modifications.
Ingestion points: The skill performs systematic audits of files in packages/ and apps/ directories using shell commands like find and grep (found in SKILL.md and references/sweep-procedure.md).
Boundary markers: The instructions do not specify boundary markers or delimiters to differentiate between instructions and data when processing the code files.
Capability inventory: The agent has the capability to modify production code and delete test files as specified in Procedure Step 5 of SKILL.md and Step 5 of references/sweep-procedure.md.
Sanitization: There is no evidence of content sanitization or validation of the files before the agent acts on the information.
[COMMAND_EXECUTION]: The skill provides shell scripts for auditing the codebase which use standard utilities.
Evidence: SKILL.md and references/sweep-procedure.md contain shell loops using find, wc, and grep to identify refactoring candidates based on Lines of Code (LOC) ratios and specific code patterns. These commands operate locally on the workspace files and do not involve network operations.

cohesion-over-testability