The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the user's codebase (code, comments, configuration) which acts as an ingestion point for potential malicious instructions.
Ingestion points: The agent is instructed to read files throughout the repository for analysis and auditing as specified in SKILL.md and references/tech-debt-audit.md.
Boundary markers: No explicit instructions are provided to wrap external content in delimiters or to ignore embedded instructions within the files being read.
Capability inventory: The skill utilizes filesystem access tools and is instructed to execute various shell commands for auditing in references/tech-debt-audit.md, providing a platform where injected instructions could manifest.
Sanitization: There is no requirement for the agent to sanitize or validate the content of the repository files before processing them.
[COMMAND_EXECUTION]: The technical debt audit process involves executing multiple shell-based development tools and utilities.
The skill instructs the agent in references/tech-debt-audit.md to use git, rg (ripgrep), and ast-grep for repository navigation and search.
It suggests running ecosystem-specific security and quality tools such as npm audit, pip-audit, ruff, mypy, and cargo audit to identify vulnerabilities and technical debt.
These operations are consistent with the skill's stated purpose and follow the practice of not installing tools globally without permission.

kernel-clean-code