PostHog Exception Triage Agent with Hermes

Reading an event trail and deciding whether it amounts to reproduction steps is ordinary comprehension applied to messy data, not a coding problem. Hermes pairs an open-model foundation with capable general reasoning and genuine tool access, which suits a job whose hardest moment is describing what a stranger was doing in the seconds before their session broke.

Loading preview…
Free to start · guided setup

Watch it work before it's live

Run a staged conversation — no account needed. The agent handles it for real while a simulated world answers its tool calls; nothing touches real accounts, and nothing is actually sent.

TypeError: t.workspaceId is undefined

user_8841 (dana.okoro@brightsidehealth.io)web app, project Production

at resolveWorkspace (assets/index-4f2a91.js:1:88214), at WorkspaceSwitcher (assets/index-4f2a91.js:1:91002), at renderWithHooks. Thrown on /settings/members three minutes into the session, with flag new-workspace-switcher active.

Set up in minutes

Using this template drops you into a guided setup. It asks exactly this, nothing else:

  1. Connect PostHog

    One sign-in. The agent acts through your account, scoped to what this template uses.

  2. Connect Linear

    One sign-in. The agent acts through your account, scoped to what this template uses.

  3. Connect Slack

    One sign-in. The agent acts through your account, scoped to what this template uses.

  4. Triage rules

    How you decide an exception matters and where it goes - the person and occurrence counts for each severity, the window you count over, the Linear team that receives issues, and the exceptions you already know about and want left alone.

  5. Runs on Hermes

    Preselected for this page — connect your Hermes account during setup, or switch to NoClick's built-in models with one click.

  6. Watch it handle a test run

    A staged conversation against a simulated world — then it’s live.

Why Hermes for this agent

Narrative from a click trail

Pages, clicks and active flags have to become a sentence an engineer can follow. That is comprehension work rather than API work.

Refusing to name a cause

The agent holds a stack, a trail and a count, none of which identify the change that introduced the bug. Leaving that blank under pressure is the behaviour to watch on any model.

Before you fork

If the issue bodies read thin, what do we change first?

The model, then the harness, and the triage rules last. Hermes runs on the model you connect, so the cheapest experiment is a stronger one against the same three staged exceptions. The section that improves or fails to is WHAT THEY WERE DOING, since turning a click trail into readable steps is the only part that genuinely needs comprehension.

Run it with a different agent

Put PostHog Exception Triage Agent to work on Hermes

Free to start. Guided setup, a test run against staged conversations, and it's live.