PostHog Adds Coding Agent to the PostHog Slack App.

The PostHog Slack app now puts a coding agent inside Slack. Tag @PostHog in a thread to question product data, investigate a bug or turn a message into a draft pull request.

The latest LinkedIn post highlights one of its suggested prompts: “Fix it, make no mistakes”. Behind that short instruction is an agent that can inspect a thread, work on code in a sandbox and return a draft PR.

It also brings PostHog’s existing data assistant into conversations. A team can ask why EU sign-ups dropped, request a customer briefing or change the rollout percentage for a feature flag without moving the discussion elsewhere.

The PostHog Slack app page describes one agent with two jobs. For analytics, it uses the same SQL-writing assistant as PostHog AI. For coding, it creates a sandboxed environment, plans the requested work, edits files, runs checks and opens a draft pull request.

That means a thread can move from identifying a product problem to preparing a code change. PostHog gives the example of asking why EU sign-ups dropped and then having the agent open the PR intended to fix the issue.

The coding agent can also add a retry policy to a webhook worker or resolve merge conflicts on an existing pull request. Work stays attached to the original conversation, so later instructions can come from other people in the thread.

This adds an immediate coding route to PostHog’s agent work, following PostHog scouts Bring Scheduled, Learning Agents to Product Analytics. Scouts run scheduled analysis, while the Slack app responds to requests made in a conversation.

The agent starts by scanning the full thread. That includes written messages and supported attachments, rather than only the message containing the @PostHog mention.

It then carries out the code work in a sandboxed environment. Tests and checks run there before the app creates a draft pull request with a detailed description and posts the link back into Slack.

Each follow-up turn runs under the sender’s own GitHub and PostHog identity. This lets several teammates steer the work while retaining the identity of the person making each request.

The agent also watches continuous integration after creating the pull request. It can rerun jobs when failures appear to come from the environment, but it does not edit workflow files. It automatically triages review comments as they arrive.

PostHog lists a range of requests spanning analytics, feature flags and code maintenance. The same workflow can also run from the Slack mobile app, allowing a request and its follow-up instructions to be sent from a phone.

The draft status keeps generated changes in the review process rather than presenting them as merged work. Teammates can inspect the description, respond in the original thread and add further instructions before the pull request progresses.

What is the PostHog Slack app?

The PostHog Slack app is an agent for product-data questions and coding tasks inside Slack. It can analyse PostHog data, work on code in a sandbox and open draft pull requests from a thread.

How does the PostHog Slack app compare with PostHog AI?

Its analytics function uses the same SQL-writing, statistically minded assistant as PostHog AI, but responds inside Slack. The app also adds a coding agent that can edit files, run checks and create draft pull requests.

Which attachments can the PostHog Slack app read?

The agent can scan images, PDFs and text files attached to a Slack thread. It uses those files alongside the thread’s messages when planning the requested work.

Can teammates update a draft pull request from the Slack thread?

Yes. Anyone in the thread can send follow-up messages to steer the run. Each turn uses that sender’s own GitHub and PostHog identity.