The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external PostHog projects, which could contain instructions intended to influence the agent's behavior.
Ingestion points: POSTHOG_LIST_AND_FILTER_PROJECT_EVENTS and POSTHOG_LIST_AND_MANAGE_PROJECT_FEATURE_FLAGS retrieve event properties and flag metadata from the PostHog API.
Boundary markers: The instructions do not define clear delimiters or "ignore instructions" directives for data retrieved from external sources.
Capability inventory: The skill has the ability to perform write actions, including capturing events (POSTHOG_CAPTURE_EVENT) and creating feature flags (POSTHOG_CREATE_FEATURE_FLAGS_FOR_PROJECT).
Sanitization: There is no mention of sanitizing or validating the properties or metadata retrieved from external sources before they are processed by the agent.
[EXTERNAL_DOWNLOADS]: The skill directs the user to connect to a remote MCP server to enable its functionality.
Endpoint: https://rube.app/mcp is the required server configuration for the Rube MCP toolkit provided by Composio.

posthog-automation