The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes extensive amounts of untrusted source code from the user's project to perform architecture, bloat, duplicate, and retry audits. Adversarial instructions embedded within code comments, documentation, or string literals could potentially manipulate the agent's audit findings or influence its recommendations.
Ingestion points: The skill reads the entire codebase using sub-agents, including scripts, configuration files, and project metadata (referenced in architecture-audit.md, bloat-audit.md, and duplicate-audit.md).
Boundary markers: There are no explicit instructions to the agent or its sub-agents to ignore or delimit embedded natural language instructions within the source code being analyzed.
Capability inventory: The skill has the ability to read and write files (to propose and apply code changes) and utilizes platform tools for searching (e.g., grep, git log).
Sanitization: The skill does not perform any sanitization or validation of the code content before it is interpolated into the agent's context for analysis.

code-upgrade