The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface due to its external data ingestion workflow.
Ingestion points: The agent is instructed to perform 'Account Research' and 'Attendee Research' in Phase 1 and 2, which involves gathering data from external sources like news websites and social media profiles (SKILL.md).
Boundary markers: No explicit delimiters or instructions are provided to the agent to isolate or ignore instructions that might be embedded in the external research data.
Capability inventory: The agent can read local files and execute an included Python validation script (scripts/check-output.py).
Sanitization: There is no logic present to sanitize or filter external content before it is processed into the final meeting brief.
[SAFE]: The Python script scripts/check-output.py is a benign validation utility that uses standard library modules (pathlib, sys) to check for required headers in output files. It does not perform network operations or access sensitive system paths.
[SAFE]: All external URLs and expert citations in references/experts.md correspond to legitimate professional platforms and official vendor domains, including those belonging to the skill author, LeadMagic.

meeting-prep