The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external code diffs and repository content, which are untrusted data sources.
Ingestion points: The agent reads source code, goals, and git diff output as part of the audit process defined in SKILL.md.
Boundary markers: There are no specific delimiters or instructions to ignore potential commands embedded within the code being reviewed.
Capability inventory: The skill guidelines do not include any tool invocations, shell commands, or network requests.
Sanitization: Input data is analyzed for semantic density rather than being sanitized or escaped.
[SAFE]: The skill contains explicit safety overrides, instructing the agent to never remove code related to security, trust-boundary validation, or accessibility during the simplification process.
[SAFE]: A reference to an external repository in the YAML frontmatter (vendored_from) is purely informational and does not initiate any remote code execution or automated downloads.

deslop