> AI agents: this is one page from PostHog's docs. Full index of Markdown docs for LLMs: https://posthog.com/llms.txt

# PostHog Desktop settings

Use Settings to customize PostHog Desktop and choose defaults for new agent sessions. Search Settings to find a specific option.

## General

Set your PostHog account, theme, and defaults for new tasks. This includes the [autonomy mode](/docs/posthog-desktop/tasks.md#autonomy-modes), effort level, composer behavior, and diff view. You can also keep your machine awake while agents run and manage [updates](/docs/posthog-desktop/download-posthog-desktop.md#updates).

## Notifications

Choose push notifications, dock behavior, completion sounds, and spoken narration.

## Personalization

Use **Custom instructions** to add instructions to every agent session. For example, write "always write tests" or "prefer functional patterns." PostHog Desktop can sync instructions from `AGENTS.md` or `CLAUDE.md` in your repository.

This section also contains hedgehog mode and sound effects.

## AI-generated content

PostHog Desktop can generate:

-   **Branch names** – Derived from your task title, so the name on disk reflects the work it contains
-   **Commit messages** – A concise imperative subject line that respects your repository's commit conventions
-   **Pull request titles and descriptions** – A summary of the actual diff, added to `PULL_REQUEST_TEMPLATE.md` when your repository has one

For agent tasks, PostHog Desktop also uses the conversation to carry the reason for a change into the commit and pull request. You can edit, regenerate, or replace any generated content.

## Plan & usage

Review AI credit spend by day and model, and set a billing limit. See [pricing](/docs/posthog-desktop/pricing.md) for prices.

## Skills and MCP servers

Manage [skills](/docs/posthog-desktop/skills.md) and [MCP servers](/docs/posthog-desktop/mcp-servers.md). MCP server settings let you allow, block, or require confirmation for tools.

## Environments

Local environments define setup scripts for new checkouts. They use TOML and can be committed to the repository.

Cloud environments set network rules and environment variables for remote sandbox sessions. See [environments](/docs/posthog-desktop/environments.md).

## Workspaces and worktrees

Manage repositories added to PostHog Desktop. Configure [checkout auto-suspension](/docs/posthog-desktop/worktrees.md) with a maximum active count and inactivity threshold. PostHog Desktop restores suspended checkouts when you open them.

## Harness

Review Claude Code and Codex configuration paths. The Permissions view shows which Claude Code operations are allowed or denied.

**Allow bypass permissions mode** adds bypass mode to the composer's mode menu. It does not turn bypass on. Sessions keep asking for approval until you select bypass for one. A session in bypass mode runs shell commands, file edits, and web requests without asking, so use it only in an environment you can restore.

## Workspace tools

Every task includes tools for checking and adjusting the agent's work without switching apps.

### Terminal

The terminal opens in the task's working directory. Local tasks use your repository, while worktree tasks use their isolated working directory. Run tests, migrations, `git log`, or a development server from a real shell. File changes appear to the agent like any other edit.

### Editor

Open files from the [diff panel](/docs/posthog-desktop/code-review.md) or repository tree and edit them in place. Markdown files have a preview. Hover a PostHog SDK call, such as `capture()` or a flag check, to see its event volume or rollout.

Select a range and leave a comment to send the selected code and your instruction to the agent. Use **Open in editor** when you prefer your external editor.

### Browser tabs

Open a web page beside a task to check a preview, dashboard, or doc without leaving PostHog Desktop. Tabs belong to the window where you open them. A [space](/docs/posthog-desktop/spaces.md) section such as `CONTEXT.md` can open as a tab too.

### Watch agents from other windows

Mission Control shows running agents over your other windows, so you can watch progress while working elsewhere. Turn it on in **Settings** > **General**.

### Spoken narration

With narration enabled, the agent speaks short status updates while it works. Turn it on in **Settings** > **Notifications**.

### Shortcuts and sidebar

Configure keyboard shortcuts and choose which items appear in the sidebar.

## Integrations

-   **GitHub**: Project and personal connections to the PostHog GitHub App. See [GitHub integration](/docs/libraries/github?tab=Desktop.md), and [Git + GitHub](/docs/posthog-desktop/git-github-integration.md) for the local Git menu
-   **Slack**: Workspaces for [starting tasks from Slack](/docs/slack.md), inbox notifications, and comment notifications. See [Slack integration](/docs/libraries/slack?tab=Desktop.md)
-   **Discord**: Optional Rich Presence status. See [Discord integration](/docs/libraries/discord?tab=Desktop.md)

## Self-driving

Configure [self-driving](/docs/self-driving.md) work without leaving PostHog Desktop:

-   Choose which [signal sources](/docs/self-driving/inbox/sources.md) feed your [inbox](/docs/self-driving/inbox.md)
-   Set the daily report limit
-   Set the priority at which the agent starts assigned tasks automatically
-   Choose a base branch for automatically opened pull requests in each repository
-   Send actionable reports to a team Slack channel or notify suggested reviewers directly

This is the same configuration used by the PostHog web app and the **Agents** tab. A change in one place applies everywhere.

See [Use Self-driving in PostHog Desktop](/docs/posthog-desktop/self-driving.md) for the report review and agent handoff workflow.

## Advanced

Use Advanced settings for troubleshooting, storage resets, developer mode, and low-level cloud or command options.

### Still have questions?

Ask PostHog AI

### Was this page useful?

HelpfulCould be better