The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it processes untrusted data from meeting transcripts to identify patterns.
Ingestion points: Meeting transcripts are ingested through the get_meetings_for_period action and subsequent API calls to /api/insights and /api/search endpoints.
Boundary markers: The skill relies on semantic search queries and extraction rules to filter relevant content; it does not explicitly define structural delimiters (like XML tags) to separate transcript content from agent instructions.
Capability inventory: The skill is granted calendar and crm capabilities, which are used for data retrieval. It lacks dangerous capabilities such as file system writing, subprocess execution, or arbitrary network access.
Sanitization: The skill includes specific logic to distinguish between prospect objections and internal team dialogue or general questions, providing a layer of semantic filtering.
[SAFE]: The skill uses platform-integrated tools and APIs for its operations.
All data retrieval and analysis are performed via execute_action or relative API paths (e.g., /api/search, /api/insights), which are part of the trusted platform environment.
There are no references to external third-party domains, hardcoded credentials, or unauthorized data exfiltration patterns.

Meeting Objection Tracker