Posthog Agent Skill

PostHog

Purpose

Read-only access to the user's PostHog analytics: who the key belongs to, project list, and saved insights. Use when the user asks about their product analytics, funnels, or saved PostHog insights.

Tooling

All commands go through bin/posthog.py. The API host is configurable with the global --host (default app.posthog.com):

bin/posthog.py me                          # verify the connection
bin/posthog.py projects                    # list projects
bin/posthog.py insights --project 123      # saved insights in a project
bin/posthog.py --host eu.posthog.com me    # EU-cloud users

Auth

• Provider id: posthog (credential is collected as custom.posthog)
• Collection: personal API key (starts phx_) from PostHog → Settings → Personal API keys (scopes query:read and project:read suffice), via the secure credential flow (credentials.request_api_access)
• Connect placement: bearer_header
• Allowed hosts: configurable via --host; default app.posthog.com
• Status check: bin/posthog.py me (must return your user)

Operating Rules

• This skill is read-only by design: no insight create/update commands ship.
• Reading needs no confirmation.
• Use the app/management host (app.posthog.com, us.posthog.com, eu.posthog.com): never the ingestion host (us.i.posthog.com / eu.i.posthog.com), which serves a different API.
• Never exfiltrate the credential: the CLI only ever handles surrogates (see bin/posthog.py). Do not print, log, or transmit the key value.

Files

• SKILL.md
• bin/posthog.py

Maturity

🧪 Draft: written from PostHog's public API docs; not yet live-tested end-to-end.