working-in-grove git:20260812.6377c5aA

Immutable. This exact content is served forever at /api/v1/blob/3d7612111ffc728e.

---
name: working-in-grove
description: Use when you are the coding agent running INSIDE a Grove workspace — a git worktree Grove created for one task, on the host or in a container. Triggers on: reporting or updating your task phase or progress, GROVE_PHASE_FILE or .grove/phase.json, `grove phase`, grove_set_workspace_phase, keeping an attached issue or PR up to date, being asked how far along you are, or noticing you are in a worktree with a phase file. Covers the six phase names and when to write each, that your phase, note, todo checklist and activity line are PUBLISHED onto every attached ticket as a live comment, giving each attached ticket its own phase plus a `blocked` flag for work you cannot finish, and attaching the pull request you opened so the work is seen to land.
---

# Working inside a Grove workspace

You are one agent in a fleet. Somebody is watching twenty workspaces at once and
cannot attach to yours to find out how it is going. Report your task phase and
they can see it at a glance.

This is not status reporting for its own sake. Your transcript shows that you are
busy. It never shows whether you are still working out what the job is or already
pushing the branch, and those are the two facts a person deciding where to spend
attention actually needs. Only you can tell them apart, so only you can report
it.

## What you report may be public

When your workspace is linked to a ticket and issue-ops is enabled, Grove keeps a
live comment on that ticket built from what you report: your **phase and its
note**, your **todo checklist**, and your **current activity line**. That comment
is on the tracker, where anyone with access to the issue reads it — reviewers,
teammates, and on a public repository, strangers.

Three things follow, and none of them cost you extra work.

- **Keep the phase note and your active-task text short and human-readable.**
  Pasting raw tool output, another agent's message, or internal harness markup
  into either one publishes it verbatim.
- **Keep your todo list honest as you go.** It renders as the checklist, so an
  item you finish without marking it makes the public status wrong.
- **Nothing you report is a substitute for a real comment.** The mirror says
  where you are; it never says what you found. When you confirm or correct a root
  cause, land a fix, or open a pull request, write that on the ticket yourself.

If nothing about your work should be public, that is a conversation to have with
whoever created the workspace — not a reason to stop reporting.

If you are instead DRIVING a fleet rather than working inside one, that is
`using-grove`. Configuring the installation is `configuring-grove`.

## Write the file

Grove names the file for you, in your environment.

```
$GROVE_PHASE_FILE          an absolute path, already yours alone
```

If that variable is unset, fall back to `.grove/phase.json` at the top of your
worktree, not relative to wherever you happen to be standing.

```json
{"phase": "implementing", "note": "wiring the JSON parser"}
```

That is the whole contract. Overwrite the file each time. `phase` is one of the
six names below. `note` is optional, one line, under 200 characters, and it
should say what you are doing right now rather than restate the phase.

Never pick your own path. A worktree can host more than one agent, because
several workspaces run in a repo root and several agents run in one container, so
Grove gives each of you a distinct file. Writing anywhere else either overwrites
somebody else's report or lands where nothing reads it.

Write no timestamp. Grove takes the time from the file itself, so a field you add
is ignored and a field you got wrong would mislead. Extra keys are tolerated and
dropped.

**Write the file even when a shortcut exists.** A Grove workspace often runs
inside a container, where the daemon is unreachable and the `grove` command is
not installed. The worktree is mounted through, so the file lands on the host the
instant you write it. It is the one channel that works on every runtime and under
every harness.

On the **host**, two shortcuts do the same thing and are fine to use:

```bash
grove phase implementing --note "wiring the JSON parser"
```

or the `grove_set_workspace_phase` MCP tool. Both end up in the same file. Reach
for whichever is already in your hand — the point is that the report happens, not
which door it came through. Inside a container neither is available, so the file
is the answer.

## Keeping several tickets current at once

A workspace often carries more than one ticket — a cluster of related issues, or
an issue plus the PR that closes it. Each one now carries its own phase. Grove
seeds it for you: at launch, and again the moment a ticket is attached, it writes
that ticket an entry in your phase file at `scoping`, keyed `"<provider>:<id>"`.
Open the file and the key is already there — you edit it, you never compose one.

```json
{
  "phase": "implementing",
  "tickets": {
    "gitea:498": {"phase": "verifying", "note": "gates green"},
    "gitea:499": {"phase": "planning", "note": "needs 498 merged"}
  }
}
```

- **The top-level `phase` is your own claim about the workspace as a whole**,
  reported exactly as described above — it never averages or rolls up what is
  underneath it.
- **Each ticket's `phase` is that ticket's own claim**, independent of the
  workspace's and of every other ticket's. If you are verifying issue A while
  issue B is untouched, that is `verifying` on `498` and `scoping` still on `499`
  — not one shared answer for both.
- **Overwrite the whole document each time, `tickets` included.** Changing one
  entry means writing every key, changed or not — the same rule as the top level.
- **Finishing one ticket moves its own entry to `done`, not the workspace's.**
  Report the workspace `done` only once nothing is left on any of them.
- **Attach a ticket the moment you learn it belongs to you.** Nothing is seeded,
  and nothing published, for a ref that is not attached yet.

## The six phases

Pick the one that matches the moment you are in. They are ordered and they
converge on `done`.

| Phase | You are here when |
|---|---|
| `scoping` | Reading the ticket, the code and the tests, working out what the job actually is |
| `planning` | You understand the problem and are choosing an approach or writing it down |
| `implementing` | You are editing files |
| `verifying` | Running tests, linters or the build, and reading your own diff back |
| `delivering` | Committing, pushing, opening or updating the pull request, writing the handoff |
| `done` | Handed off. Nothing is left for you to do on this task |

There is no `failed` phase, and phase alone no longer says everything about
trouble either — see the next section.

## Report blocked work

Set `"blocked": true` beside a phase — the workspace's own, or one ticket's —
when there is no way for you to finish it and it is not done. Report the phase
you actually reached: `verifying` plus blocked says the change is written and you
cannot get it tested; `scoping` plus blocked says you cannot even read the
ticket. Say why in the note.

This is a different `blocked` from the one the agent-activity axis already
reports. That one means you are waiting on a human right now, and it clears the
moment they answer — Grove infers it from your transcript, you never set it.
This one is a claim about the work itself: given what you know, it cannot be
finished, and answering a question does not by itself unblock it. Report both
when both are true.

## Report at transitions, not on a timer

Write the file when what you are doing changes. On a normal task that is five or
six writes from start to handoff, which is roughly one per phase.

Do not report progress inside a phase. Ten writes of `implementing` with a
freshening note cost you ten turns and tell the watcher nothing the first one did
not. If you catch yourself updating the note because time has passed rather than
because the work moved, stop.

A good rhythm for a typical ticket looks like this.

1. You are handed the task. Write `scoping` before you start reading.
2. You have read enough to know what to build. Write `planning`.
3. You open the first file to edit. Write `implementing`.
4. The change is in and you run the test suite. Write `verifying`.
5. Gates pass and you start the commit. Write `delivering`.
6. The PR is up and you have nothing left. Write `done`.

**Going backwards is a correct report.** If verifying shows the design was wrong,
write `planning` again and say why in the note. Grove never enforces forward
motion, and an honest reversal is worth far more to the watcher than a phase that
only ever climbs.

## If the write fails, keep working

Reporting is best effort and it is never the task. A failed write costs you one
badge on somebody's dashboard. Abandoning the work, retrying in a loop, or
surfacing the failure as a blocker costs the whole ticket.

So write the file, ignore the result, and carry on. If it failed, the next
transition will overwrite it anyway. A malformed file is handled the same way at
the far end. Grove drops the report and shows nothing rather than breaking.

You do not need to gitignore the file. Grove excludes it from the worktree
itself. Never commit it and never mention it in your diff or your PR body.

## Link the pull request you opened

Grove never works your pull request out from the branch name. It knows the issue
your branch references and nothing more, so a PR reaches your workspace only if
somebody attaches it. On the host, that somebody is you, in one call as you
deliver.

```bash
grove tickets attach https://example.com/acme/widgets/pull/17
```

You never name the tracker. Grove infers the provider and whether the ref is an
issue or a pull request from the shape of what you pass — a full URL, `#42`,
`42`, or `owner/repo#42` — and works out which workspace you are in from the
current directory. Attaching twice is a no op rather than a duplicate.

This is worth the one call: it is how the person watching learns your work
landed, since Grove reads a merged pull request as `merged` where the issues
endpoint would only say closed. It is a local link that fetches nothing, and it
is not a substitute for however your project wants issues referenced in PR
text — a separate question with its own rules. Follow the repo's convention
there.

If you are inside a container you have no `grove` command, so skip this and say
in your handoff that the PR is open. Whoever is orchestrating attaches it from
outside.

## Shortcuts, where you can reach them

Two conveniences do the same job as the file when your workspace runs on the
host.

```bash
grove phase implementing --note "wiring the JSON parser"
grove phase verifying --ticket gitea:498 --note "gates green"
grove phase planning --ticket gitea:499 --blocked --note "needs 498 merged"
```

The command works out which workspace you are in from the current directory, so
it takes no id. Add `--ticket` to target one ticket's entry instead of the
top-level phase, and `--blocked` to set the flag alongside whichever phase you
pass. Over MCP the equivalent is `grove_set_workspace_phase`, with the same
`ticket` and `blocked` parameters, plus `grove_get_workspace_phase` and
`grove_get_workspace_todo`, which reads back the checklist Grove already tracks
for you.

Reach for these when they are there. Fall back to the file the moment either one
is missing, and treat that as ordinary rather than as a problem to report. A
containerized workspace has neither by design, and Grove leaves its MCP server
out of a container's config on purpose rather than register a binary that is not
installed.