The Agent Skills Directory

[SAFE]: The skill serves as a technical guide for an AI agent to query the document_embeddings table using PostHog's HogQL. The instructions are focused on legitimate data retrieval and analysis tasks.
[DATA_EXFILTRATION]: While the skill enables the retrieval of potentially sensitive data (content and embeddings), it does so through the authorized posthog:execute-sql tool within the user's own environment. There are no instructions to transmit this data to external or unauthorized destinations.
[COMMAND_EXECUTION]: The skill documents the use of the posthog:execute-sql MCP tool. It emphasizes mandatory filters (model_name, product, document_type, and timestamp) that ensure query performance and data relevance. It also notes that the HogQL engine automatically enforces team-level data isolation.
[INDIRECT_PROMPT_INJECTION]: The skill defines an attack surface by instructing the agent to ingest natural-language descriptions from a database. However, this is the core functionality of a data-querying skill.
Ingestion points: Content is retrieved from the document_embeddings table via HogQL queries (SKILL.md).
Boundary markers: None explicitly defined in the SQL templates, though the documentation focuses on structured data extraction.
Capability inventory: The skill utilizes the posthog:execute-sql tool for database access.
Sanitization: No specific sanitization of retrieved database content is mentioned, which is standard for internal data-analysis tools.

signals