Control PostHog credential retrieval for AI agents
The PostHog preset stores a personal API key and an optional host. Agents can retrieve approved fields before calling PostHog from their runtime. Use the host for your US, EU or self-hosted installation, and do not confuse a personal API key with the project token used for event capture.
Separate analytics access from event ingestion
A reporting agent may need access to product data or configuration that an event-capture token does not provide. Check the personal key's scopes and project access in PostHog, and choose only the permissions required for the workflow. Gazebo adds named retrieval grants and access records; it does not inspect query results, constrain analytics requests after token delivery or replace PostHog's own permissions. Product analytics can contain sensitive customer information even when access is read-only.
How it works
- 1
Create a PostHog personal API key with the intended project access and minimum scopes for the task.
- 2
Choose PostHog in Gazebo and store the personal_api_key, plus the host when your runtime needs it.
- 3
Grant retrieval to the relevant agent. Its runtime selects the correct project and API endpoint and calls PostHog directly.
- 4
Review Gazebo credential-retrieval history separately from PostHog activity. Revoke or rotate retained keys at the provider when needed.
Common use cases
Recurring product reports
Let a reporting agent retrieve a dedicated personal API key for the intended project. Review the provider permissions and query output handling rather than assuming a reporting task cannot expose customer data.
Cloud and self-hosted installations
Store the appropriate host alongside the key so the runtime can select the correct installation. Keep different installations' credentials and agent grants separate.
Connect PostHog to Gazebo
Control agent credential retrieval for PostHog. Record retrieval attempts and revoke future access.
Connect PostHog