The Agent Skills Directory

[SAFE]: The skill implements standard data warehouse management functionality using official posthog namespace tools. It identifies gaps in native data collection and suggests appropriate external sources (Stripe, HubSpot, etc.) to bridge them.
[DATA_EXPOSURE]: The skill uses posthog:external-data-sources-list and posthog:read-data-warehouse-schema to read metadata about existing data connections. This access is scoped to the agent's authorized tools and is necessary for the stated purpose of guiding data imports.
[COMMAND_EXECUTION]: The skill uses posthog:execute-sql to demonstrate data joining capabilities. This is a controlled execution environment within the PostHog platform for data analysis.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process and join data from external third-party systems like CRMs and billing platforms.
Ingestion points: Data enters via external schemas and data warehouse tables retrieved through posthog:external-data-schemas-list and posthog:read-data-warehouse-schema (SKILL.md).
Boundary markers: None explicitly defined in the instructions for separating external data from instructions.
Capability inventory: The agent can execute SQL queries via posthog:execute-sql (SKILL.md).
Sanitization: The skill does not specify sanitization or escaping for the external data it instructs the agent to query.
Note: This represents a standard surface for indirect injection inherent to data integration tools, rather than a malicious pattern.

suggesting-data-imports