How do approvals work?

By default, every change an agent makes in a connected app waits for your approval, on every plan. The run pauses and the proposed action lands in the approvals inbox. Approve it and the run continues; reject it and the agent is told why. On each connection you choose, tool by tool, whether it asks, runs on its own (destructive calls still ask) or is switched off. Enhanced Control adds per-app read, write and destructive rules at the gateway on Free and Pro.

Can I use PostHog from Claude, ChatGPT or Cursor?

Yes. Point any MCP client at your Spinrun endpoint and it can use PostHog through the same connection your Spinrun agents use, with every call in the same activity log.

What happens when I connect a second PostHog account?

You give each connection a title and a note. From then on every PostHog tool name ends in a short form of the title, and the note joins the tool's description, so the model knows what each account holds. Risk tiers and the activity log still follow the tool.

Free includes 10,000 one-time starter credits, no card required. Pro starts at €20 a month for 20,000 shared credits and adds manual top-ups. Both include the same product features. Credits meter the work agents do, not the number of people supervising them.

PostHog for AI agents

Product analytics, feature flags, session replay, and experiments. Add an API key once and every agent you run can use it.

Connect PostHog All toolkits

Connects with an API key
Ready to connect
Any MCP client
No card required

The route from an agent to PostHog

Once PostHog is set up in your workspace, any agent you run can reach it: a Spinrun agent, or Claude, ChatGPT and Cursor through one MCP endpoint. Every call passes through the Spinrun gateway, which sorts it by risk and writes it to the activity log.

By default, every change an agent makes in a connected app waits for your approval, on every plan. Enhanced Control adds per-app read, write and destructive rules at the gateway on Free and Pro.

PostHog in the catalog

Product analytics, feature flags, session replay, and experiments. · Analytics & Data

An agent gets a jobFrom chat, a schedule, or your own AI client
Spinrun checks the callSorted as a read, a write or a destructive call
PostHog does the workThrough the connection you set up in your workspace
Every call is loggedTool, outcome and duration, in one activity log

From Spinrun recipes

Jobs to hand an agent in PostHog

Each one is a real prompt from a Spinrun recipe or team page, ready to adapt to your own work.

“What changed in the funnel last week? Two sentences on the Growth page in Notion, backed by Analytics and PostHog.”
Weekly funnel note
“Sprint planning on one Notion page: which items in the Linear backlog does PostHog's funnel from last week argue for?”
Sprint planning page
“Did people stick with what we launched last Tuesday? Tell #product what Mixpanel and PostHog show for retention and adoption.”
After-launch recap

Three steps

From connection to first call

Connect an app once
Sign in from your workspace with OAuth or an API key. Spinrun keeps the credential in an encrypted vault, never in an agent's config.
GMAIL_FETCH_EMAILSDone
HUBSPOT_RETRIEVE_DEAL_BY_IDDone
GMAIL_SEND_DRAFT
Waiting for your approval
ApproveReject
Decide what it may do
By default, every change an agent makes in a connected app waits for your approval, on every plan. Enhanced Control adds per-app rules for reads, writes and destructive calls on Free and Pro.
Activity log
ClaudeNOTION_SEARCH_NOTION_PAGEDone
CursorLINEAR_CREATE_LINEAR_ISSUEDone
ChatGPTGMAIL_FETCH_EMAILSDone
Spinrun agentHUBSPOT_RETRIEVE_DEAL_BY_IDDone
Let any agent use it
Claude, ChatGPT, Cursor or a Spinrun agent reach the app through one MCP endpoint, and every call is written to the same activity log.

What it can do

Every PostHog action, sorted by risk

Spinrun sorts every PostHog action into one of three tiers. By default, agents ask before a write; on Free and Pro, the rules you set decide which tiers run.

Writes
Add dashboard collaborators with access level
POSTHOG_ADD_DASHBOARD_COLLABORATORS_WITH_ACCESS_LEVEL
Add collaborators to a specific project dashboard. requires 'project id' and 'dashboard id'. supports json, form-urlencoded, and multipart data. returns added collaborator details. auth: 'personalapikeyauth'.
Writes
Add member to organization role
POSTHOG_ADD_MEMBER_TO_ORGANIZATION_ROLE
Add a member to an organization role by submitting their details and uuid in json/form, using organization and role ids. requires personalapikeyauth. success gives a 201 status.
Writes
Add project specific member roles
POSTHOG_ADD_PROJECT_SPECIFIC_MEMBER_ROLES
Add members with specific roles to a project by providing their user uuid and level (member or admin). requires project id in the url. supports json, form-data, and form-urlencoded input. returns the newly created membership data upon success.
Writes
Assign feature flag role based access
POSTHOG_ASSIGN_FEATURE_FLAG_ROLE_BASED_ACCESS
Set up role-based access for a feature flag using `project id` and `feature flag id`. submit access info in json or form format, with `feature flag:write` rights. success returns new details with a 201 code.
Writes
Bookmark project activity notification
POSTHOG_BOOKMARK_PROJECT_ACTIVITY_NOTIFICATION
Create a bookmark for an activity notification in a project's activity log. requires a `project id` and json payload with activity details. supports verification status, impersonation flags, and system activity distinction.
Destructive
Bulk create organization invites
POSTHOG_BULK_CREATE_ORGANIZATION_INVITES
Post /api/organizations/{organization id}/invites/bulk/: bulk-create organization invites with customizable levels, sender details, and validation checks. supports email automation and private project access settings.
Destructive
Cancel insights for a project
POSTHOG_CANCEL_INSIGHTS_FOR_A_PROJECT
The "/api/projects/{project id}/insights/cancel/" endpoint allows users to cancel insights for a project by providing a project id and insight details in json or csv, and confirms cancellation in the response.
Reads
Check pipeline destination updates
POSTHOG_CHECK_PIPELINE_DESTINATION_UPDATES
Get the newest updates and plugin info for a pipeline destination by organization and destination id via get request to /api/.../{id}/check for updates.
Reads
Check pipeline import app updates
POSTHOG_CHECK_PIPELINE_IMPORT_APP_UPDATES
Retrieve the latest update information for a specific pipeline import app by its id within an organization. this endpoint returns details such as plugin type, name, description, and version data. requires 'id' and 'organization id' as path parameters.