The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local shell script (scripts/audit_skill.sh) to perform PII scanning and validation on files within a user-provided directory. It also generates shell commands for Git operations and package management.
[DATA_EXFILTRATION]: The skill accesses and reads content from sensitive local paths such as ~/.claude/skills/ to identify hardcoded credentials, emails, and other PII. While this is the intended functionality, it involves handling sensitive user data.
[INDIRECT_PROMPT_INJECTION]: The skill processes content from external files which may contain instructions targeting the agent during the audit and generalization phases. * Ingestion points: Reads SKILL.md and referenced files recursively from the user-specified source skill path. * Boundary markers: No explicit delimiters or boundary markers are mentioned when the agent reads or edits external file content. * Capability inventory: The skill utilizes Read, Write, Edit, and Bash tools throughout its multi-phase pipeline. * Sanitization: There is no evidence of sanitization, escaping, or filtering of the ingested file content before it is processed by the agent.

publish-skill