The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted source code, which serves as a potential vector for indirect prompt injection.\n- Ingestion points: Code repositories and pull requests accessed through 'Read', 'Grep', and 'Glob' tools (defined in 'SKILL.md').\n- Boundary markers: The instructions do not define specific delimiters (such as XML tags or unique markers) to isolate ingested code from the agent's internal reasoning context.\n- Capability inventory: The skill's potential impact is significantly mitigated by its restriction to read-only tools ('Read', 'Grep', 'Glob'), as it lacks file-write or execution privileges in its 'allowed-tools' configuration.\n- Sanitization: No input sanitization, filtering, or escaping of the analyzed source code is implemented.

code-reviewer