PostHog
Arcade.dev LLM tools for PostHog analytics
2.0.1PostHog Toolkit for Arcade
Enables LLM-powered interaction with PostHog (opens in new tab) analytics — querying events, managing feature flags, experiments, surveys, dashboards, insights, and error tracking via the PostHog API.
Capabilities
- Event & data exploration: List event definitions and properties, run HogQL queries, fetch time-series trends, funnels, and cohort retention data, and compare metrics across date ranges.
- Dashboards & insights: Create, read, update, and delete dashboards; create, pin, update, and soft-delete saved insights.
- Feature flags: Create, read, update, and soft-delete flags; control rollout rules and targeting; validate key uniqueness before creation.
- A/B experiments: Create draft experiments with linked feature flags, launch and archive them, retrieve per-variant statistical results (chance to win, credible intervals, exposures).
- Surveys: Create, read, update, archive, and list surveys; review activity logs of survey changes.
- Error tracking: List and retrieve error issues with status, severity, and assignee detail.
Secrets
This toolkit requires two secrets. Set them via the Arcade secrets dashboard (opens in new tab) or through the API.
-
POSTHOG_PERSONAL_API_KEY— A PostHog Personal API Key used to authenticate all API requests. Generate one in your PostHog account under Settings → Personal API keys (top-right user menu → Your profile → Personal API keys). The key must have access to the project(s) you intend to query. See PostHog docs on personal API keys (opens in new tab) for details on creating and scoping them. -
POSTHOG_SERVER_URL— The base URL of your PostHog instance. For PostHog Cloud US this ishttps://us.posthog.com; for EU Cloud it ishttps://eu.posthog.com; for self-hosted deployments, use your own domain (e.g.,https://posthog.example.com). This tells the toolkit which instance to route requests to.
For general guidance on configuring secrets in Arcade, see the Arcade secrets documentation (opens in new tab).
Available tools(38)
| Tool name | Description | Secrets | |
|---|---|---|---|
Pin an existing insight as a tile on a dashboard. | 1 | ||
Archive an experiment by ID; PostHog can restore an archived experiment.
Archiving hides the experiment from the default list and leaves its
feature flag as it is. | 1 | ||
Archive a survey by ID; PostHog can restore an archived survey. | 2 | ||
Compare the same metric across two date ranges side-by-side. | 1 | ||
Create a new empty dashboard. Insights can be pinned to it as tiles afterward. | 1 | ||
Create an A/B test experiment as a draft, with its feature flag.
The flag stays off and nobody is enrolled until the experiment is
launched. Before creating, verify the feature_flag_key is not
already used by another experiment and that the metrics' event
names exist. | 2 | ||
Create a new feature flag.
Verify the key is not already in use to avoid duplicates. After
creation, consult PostHog SDK documentation for the user's
language/framework to integrate the flag. | 1 | ||
Save a tested query as a reusable insight.
Run the query first and confirm it returns the expected results. | 1 | ||
Create a new survey.
Verify the name is not already in use to avoid duplicates. After
creation, consult PostHog SDK documentation to integrate the
survey into the user's application. | 2 | ||
Delete a dashboard by ID. This toolkit cannot restore a deleted dashboard. | 2 | ||
Soft-delete a feature flag by numeric ID or key. Provide one of flag_id or flag_key.
PostHog refuses while a running experiment or an early access
feature uses the flag; a draft, ended, or archived experiment does
not block deletion. | 1 | ||
Soft-delete an insight by ID. | 2 | ||
Get a dashboard's full configuration including all insight tiles.
Provide dashboard_id or dashboard_name. | 1 | ||
Get an error tracking issue: status, severity, name, description, first seen, assignee. | 2 | ||
Get a time-series trend for one or more events over a date range.
Returns one series per event, and one per breakdown value when a breakdown property is
set. Unknown event names return zero counts rather than errors. | 1 | ||
Get an experiment's full configuration.
Includes status, variants, rollout, and metrics. Provide either
the numeric ID or the name. | 1 | ||
Get a launched experiment's results: exposures per variant and each metric's statistics.
Each metric reports the baseline and every other variant with its
sample count, chance to win, and credible interval. A metric whose
query failed carries an error instead. | 1 | ||
Get a feature flag's full definition including rollout rules.
Provide either the numeric ID or the flag key. | 2 | ||
Build a multi-step conversion funnel with optional property breakdowns. | 1 | ||
Get an insight's full configuration and current query results.
Provide either the numeric ID or the name. | 1 | ||
Get cohort retention data showing what percentage of users return.
Shows return rates after a user's initial event. | 1 | ||
Get a survey's full configuration.
Includes questions, targeting rules, and scheduling. Provide
either the numeric ID or the name. | 2 | ||
List all dashboards in the project.
Returns summaries (id, name, pinned, tags, created_at). Use a
dashboard's ID to get its full configuration and insight tiles. | 2 | ||
List error tracking issues.
Returns each issue's id, status, severity, name, description,
first_seen, and assignee. Use an issue's id to get it on its own. | 1 | ||
List all tracked event names in the project.
This is the discovery starting point -- use it to find valid
event names, then look up an event's properties to discover
available filters and breakdowns. | 1 |
Last updated on
Overview
Browse Arcade's catalog of pre-built MCP servers — integrations for GitHub, Slack, Google Workspace, Salesforce, and dozens of other services, ready to use with any MCP-compatible client via the Arcade Engine.
Nimble
Search, extract, crawl, and map the web. Nimble (Nimbleway) is available in Arcade as a Partner MCP server.