An agent that reads your PostHog for you. Each run it queries your project through PostHog's hosted MCP server, compares yesterday against the trailing 7-day average, and delivers a short usage report:
412 active users (+9% vs 7-day avg), 18.2k events (+4%)
Top events: pageview 9.1k (390 users), dashboard_viewed 2.4k (210), query_run 1.9k (140), invite_sent 310 (95), export_clicked 220 (60)
Callouts: invite_sent up 2.3x, almost all from one org; looks like a team onboarding, not a trend.
Reply to it wherever you use it (your own app, Slack, Discord, or the Omnara console) to drill into any number ("why did signups spike?"). It's read-only against PostHog.
- An Omnara account (app.omnara.com)
- A PostHog personal API key created with the MCP Server preset, which scopes it to one PostHog project. US and EU accounts both work, and MCP calls are free.
Open Claude Code, Codex, Cursor, or any other coding agent and say:
Read https://raw.githubusercontent.com/omnara-ai/omnara/main/examples/posthog-analytics-agent/SKILL.md and follow it to deploy the PostHog analytics agent.
Your coding agent follows SKILL.md, which is plain steps with the
exact npx omnara commands if you'd rather run them yourself. Run it again
any time to change what it reports; it updates the agent in place.
agent.yaml is the whole agent, and it needs no machine pool: Omnara calls PostHog's MCP server directly, sending the key as the bearer token on each request, so the key never appears in the config or the event log. The MCP URL limits the agent to PostHog's query tools and turns on PostHog's read-only mode, so it can't change anything in your project. Each run makes two standing queries plus at most a few follow-ups, and "steady day" is a valid result. The metrics, the call budget, and the report format are plain instruction text you can read and edit.