The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves processing external, potentially untrusted data which serves as an attack surface for indirect prompt injection.\n- Ingestion points: The skill ingests untrusted data through email triage processes (SKILL.md Step 5) and pre-meeting briefing preparation (SKILL.md Step 4), where it evaluates messages and attendee backgrounds.\n- Boundary markers: There are no explicit instructions defining the use of delimiters or warnings for the agent to ignore instructions embedded within external data in SKILL.md or the templates.\n- Capability inventory: The skill is designed to generate markdown files and interact with other tools like email managers and Notion builders (README.md), creating a pathway where malicious data could influence agent actions across tools.\n- Sanitization: The instructions do not include requirements for sanitizing or escaping external content before incorporating it into final outputs or briefing packs.

executive-assistant