# Share Goose's Output — Add Comma as an Extension

Canonical: https://commareports.com/agents/share-goose-output
Published: 2026-09-05

> Goose's extensions are MCP servers, so publishing is just another one. Add Comma, add a recipe instruction, and Goose ends a session with a link instead of a terminal buffer.

# Share what Goose produced

Goose runs on your machine with real tools, which means it does the kind of
work whose output matters to people who are not you: it reads the logs, runs
the queries, ties the incident together, and writes it up.

That write-up ends in a terminal. The incident review happens tomorrow morning,
in a meeting, from a document nobody has.

## Publishing is just another extension

Goose's extension model is MCP, so Comma is not a special case — it is one more
extension with a URL and a token from [API tokens](/docs/api-tokens). Run
`goose configure`, add a remote extension, point it at Comma's MCP endpoint.
The shape is in [MCP setup](/mcp).

## Put it in the recipe

Recipes are where this pays off, because a recipe is the thing that runs when
you are not there:

```yaml
instructions: |
  Investigate the alert, gather the evidence, and write the incident summary.
  Publish the summary to Comma as an HTML report and return only the URL.
  If a report id is already recorded for this incident, update it instead of
  creating a new one.
```

`goose run` on a schedule then produces one link per incident, and the link is
the same one the follow-up updates. See [routines](/docs/routines) if you would
rather Comma own the schedule than cron.

## Why a page

- **It renders.** Log excerpts stay monospaced, the timeline stays a table.
- **The audience is wider than your terminal.** Incident reviews involve
  support, the account team, sometimes the customer.
- **Comments anchor to the moment in the timeline** people disagree about —
  see [commenting on HTML](/comment-on-html).
- **Goose can read the replies** through the same extension and revise — see
  [letting an agent respond to comments](/agents/let-an-agent-respond-to-comments).

## Worth knowing

- **A local agent with real tools deserves a narrow token.** `reports:write`
  and nothing else — see
  [scoped tokens](/agents/scoped-tokens-for-ai-agents).
- **5 MB per report body.**
- **Reports are snapshots.** No re-execution, bounded cost.

## Try it

Free — unlimited reports, commenters and revisions.

**[See MCP setup →](/mcp)**

### Related

- [Where should my agent post?](/agents/where-should-my-agent-post) ·
  [Cron for AI agents](/features/routines/cron-for-ai-agents)
- [Share Claude Code output](/agents/share-claude-code-output) ·
  [Share Aider output](/agents/share-aider-output)
- [Routines](/docs/routines) · [The API](/docs/api)