The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external source code for OWASP security reviews, creating a surface for indirect prompt injection attacks where malicious code comments or instructions could influence the agent.
Ingestion points: File ingestion occurs via the Read, Grep, and Glob tools defined in the allowed-tools metadata of SKILL.md.
Boundary markers: There are no explicit delimiters or boundary instructions present in the skill definition to isolate untrusted code from instructions.
Capability inventory: The skill has permissions to read the local filesystem and execute git commands through the Bash tool.
Sanitization: No sanitization or input validation logic is present in this entry-point file to mitigate instruction injection from analyzed source code.

codex-security