vinzenz/hyperhive
vinzenz/hyperhive
Watch
0
0
Fork
You've already forked hyperhive
 0
Code Activity
hyperhive/docs/swarm/ui.md
iris a2e8a2bda0 swarm-ui: dropdowns open a themed list of their own instead of the browser's 
SelectField and the refresh-interval picker are built on a new Listbox, a
select-only combobox: a button that keeps focus and a listbox in the top
layer, so it works the same in every current browser, is not clipped by a
scrolling ancestor and shows above a modal dialog. Keys, type-ahead and
placement are pure functions with tests. required becomes aria-required, as a
button has no native form validation.
2026-10-25 17:15:04 +01:00

860 lines
52 KiB
Markdown
Raw Blame History

# Swarm UI
The swarm's own web surface and the operator's day-to-day view: served by
the gateway on the **swarm apex** (`services.hyperhive.swarm.domain`),
readable only by operators. The per-hive dashboard, on each hive's own
domain, covers host-level detail for one hive.
## What it shows
The signed-in user's avatar sits at the right end of the header on every page.
It opens a menu with the user's name and links to authelia's settings and
sign-out pages; the hive dashboard and the per-agent page have the same menu.
Every dropdown (a form's select, the refresh interval, the history window) opens a
list in the swarm UI's own theme, not the browser's, and works the same in every
current browser. Open it with a click, or with the arrow keys, **Enter** or
**Space** while it has focus. Then the arrow keys, **Home**, **End**, **Page Up**
and **Page Down** move through the options, and **Enter**, **Space** or a click take
the highlighted one. **Escape**, **Tab** or a click elsewhere shut the list without
changing anything. Typing the start of an option's name picks it, also with the list
shut. A dropdown inside a dialog opens above the dialog and the dialog doesn't cut it off.
| route | what |
| ---------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `/` | the start page: the swarm's name over a digital-rain backdrop, plus a paperclip offering to set up an agent while no `ruth` exists (it opens the create form with `ruth` filled in); the swarm name in the header links here |
| `/hives` | the hive directory, each hive with its last reported status |
| `/agents` | every agent: status, config PR, wanted state, inference settings, resource caps, permissions; create agents (with their permissions, optionally from an [archetype](archetypes.md)), link forge, matrix and GitHub accounts |
| `/agents/<name>/terminal` | one agent's live terminal |
| `/agents/<name>/subagents/<subagent>/terminal` | one of that agent's subagents' live terminal |
| `/alerts` | the alerts the swarm's Grafana holds now: the firing ones, with a button for the pending, no-data and error ones and a link to Grafana's own alert list; read through the controller with a Viewer service-account token |
| `/jobs` | the controller's job graph — where agent creation and credential mints show progress |
| `/schedules` | the scheduled prompts: add, edit, pause, resume and remove |
| `/messages` | the swarm-wide message feed: every message between agents, and to them from the operator, reminders and compaction, with a history window and a filter on sender or recipient; a folded **broadcast** card above the feed sends one message to several agents at once |
| `/notifications` | the notification router's flows: a switch for each built-in, every flow drawn as a graph, and an editor for custom flows |
| `/issues` | a cross-repo issue report |
| `/review` | the review classifier: its recent runs, a try-out on a real PR, and its settings |
| `/builds` | the swarm builder's recent builds of agent config PRs, with what nix printed when one failed |
Everything it shows comes from [`swarm-controller`](../../swarm-controller/README.md).
Pages that refresh on a timer share one polling rule: the page sends no new
request while the last one is still running, so an older answer never overtakes
a slow one, and a failed refresh keeps the rows already on screen under an
error.
The **wanted** badge changes the state of an agent the swarm has declared. For an
agent the swarm never declared on its hive, such as one only its hive knows
about, the badge reads **no declaration** and doesn't open: create the agent
with `swarmctl agent create` or **+ agent**, which declares it `paused`. The API
answers `PUT /api/hives/{hive}/agents/{agent}/state` for such an agent with
404 and says to create it at swarm level. Destroying such an agent needs the same
create first, since the badge is the only way to destroy.
An agent created here or with `swarmctl agent create` starts `paused`; set it
`up` from its card. A card shows **login needed** when the agent's own
turn-state header says its claude has no session, and a **log in** button
that runs the claude login from a dialog: open the URL it shows, then paste
the code back. The credentials go to the agent through the secret store
([`claude-login.md`](claude-login.md)). The agent's own web UI login still
works too.
### Agent and subagent terminals
An agent's detail panel on `/agents` shows a small live preview of its
terminal, and below it the agent's **subagents**: every subagent the swarm
queue holds terminal rows for from the last 24 hours. Picking one shows its
terminal in the same preview, and **expand** opens it in its own tab, as it
does for the agent. The list refreshes every 10 seconds, so a subagent the
agent spawns shows up once it writes output. A subagent that drops out of the
list closes its terminal, and a later subagent with the same name starts
unpicked.
The preview carries only the connection badge and **expand**, so it stays clear
of the text. What the agent is doing shows in the details table above it, in
the **state** row: the turn state (idle, thinking or compacting), a **paused**
badge for a paused agent, and the tokens in its context and billed. The row
reads "no report yet" until the first report arrives, and for an agent that's
offline. The agent sends a report when its state changes and once a minute. If
the browser gives up on the stream, a **reconnect** badge opens it again. The
terminal's own tab keeps these badges, with the model, beside the connection badge.
On a terminal's own page (`/agents/<name>/terminal`, or a subagent's) the
header has an **N running subagents** badge. It opens a list of the agent's
subagents, each a link to its terminal; on a subagent's page the current one
gets a marker and the list starts with a link back to the agent's own terminal. A
subagent counts as running when it published a row in the last 15 minutes. The
swarm doesn't report more than when the newest row came, so a subagent that
thinks quietly for longer drops out of the count. The list shows the others below,
dimmed, as **not running**.
An agent's terminal opens with its last 100 rows from the past 24 hours, then
follows new ones; a reconnect picks up after the last row it showed. A subagent
terminal does the same with that subagent's rows, and has no turn-state badges.
The browser retries a dropped connection by itself. When the swarm answers the
stream with an error status, the browser gives up and the badge reads
**disconnected**. The **reconnect** button beside it opens the stream again, and
the stream replays the rows. The messages page works the same way.
The settings menu in the header has a **hide debug output** toggle. It drops
the muted debug rows (thinking, coalesced ticks, harness chatter) from every
terminal, the preview and the tabs, and rows already on screen hide or show as
you flip it. The browser keeps the choice for the swarm UI, apart from the
per-agent pages'.
A terminal's own tab takes the agent's icon as its browser-tab icon, so tabs of
different agents tell apart. A subagent's tab shows its agent's icon. An agent
with no icon keeps the hyperhive mark.
The tab also has an agent badge, with the agent's icon and name, ahead of the
connection badge. It opens the agent's details in a popover under the badge:
the same status, state, wanted state, inference settings, resource caps, permissions and linked accounts as the
panel on `/agents`, without the terminal preview. The popover doesn't dim the
page. Selecting outside it, or Escape, closes it. From the keyboard, opening a
popover moves focus into it, and Escape puts focus back on the badge. The filter
popovers on table headers work the same way. You can stop or pause the
agent, or link an account, without leaving the terminal. A subagent's tab opens
its agent's details.
### Review
`/review` is the one page for the [review classifier](review-classifier.md),
with three sections in a column. **try a classification** and **classifier
settings** come first, each folded to a single bar: the bar of the settings
says what they're now, and what you typed in a section stays when you fold it.
**recent runs** is under them and always open. The sections share nothing and
ask for their own data.
#### Recent runs
The **recent runs** section lists what the [review classifier](review-classifier.md) has
run, newest first, one card per run. A card shows the repo and PR, the head
commit, whether the run was a try-out or a requested review, how long ago it
started, how long it took, how many classifications ran, and the outcome as a
badge. Select a card to open it: it shows the same result table as the
[try-out](#review-try-out), the overall verdict, the reason when the outcome
has one, and a link to the posted review when there is one.
An outcome can be more than approve or request changes:
| badge | what happened |
| ------------------- | -------------------------------------------------- |
| `approved` | the run posted an approval |
| `changes requested` | the run posted a request for changes |
| `comment only` | there was no clean verdict, so it posted a comment |
| `queue full` | the queue was full, so it turned the run away |
| `post failed` | the run finished but posting the review failed |
| `done` | a try-out finished |
| `failed` | the run failed; the card says why |
The page refreshes every 30 seconds, and a card you opened stays open when it
does. An empty list says there are no runs yet, and a failed load shows the reason the
controller gave. The page shows no token counts, cache counts or cost, and
stores and posts nothing.
#### Try a classification
The **try a classification** section shows how the [review classifier](review-classifier.md)
answers a classification on a real PR, before you put it in a repo's config.
Give it a repo as `owner/name` and a PR number, then write a classification: a
name, a description, a polarity (positive or negative) and a threshold from 0
to 100. **add a classification** adds more, and **run** starts the run.
The repo field offers the repos that have an open PR, and once you've picked
or typed a repo, the PR field offers its open PRs by number and title. Both
fields stay free text: type a repo or a PR the lists lack, such as a merged
one. If a list can't load, the page shows the reason and the field still takes
what you type.
**load the repo's rules** fills the classifications from the rules file on the
repo's default branch, so you can run the repo's own rules or start from them
and edit. It reads the file each time, so a rule merged a minute ago shows up.
The page names the file and the commit it read, and asks before it replaces
classifications you have written. If the repo has no rules file, the file
doesn't parse, or it holds more than ten rules, the page shows the reason and
leaves your classifications as they were.
A run takes a minute or more, and up to half an hour for ten classifications.
The page says it's running and shows how long it has been, then lists each classification with its verdict, the model's certainty, the
threshold, whether the verdict counted and the reason; the posted review folds
those fields into four columns ([`review-classifier.md`](review-classifier.md)). A classification that couldn't run shows its error where the reason
would be. Under the table are the overall verdict, approve or request changes,
and the head commit the run read. A failed run shows its error, for example when the swarm
has no API key yet.
What you typed stays after a run, so you can edit the classification and run it
again. The page saves no classification and no result, and a run posts nothing
to the forge; like every table in the UI, the result table remembers its column
filters in the browser. It shows no token counts, cache counts or cost.
When a classification answers the way you want, **open a PR** proposes the
classifications on the page to the repo. The classifier's forge account pushes
a branch that adds them to the repo's rules file, with you as the commit's
author, and the page links to the forge's page for opening the PR from that
branch, where you log in as usual. A classification named like one the file
already holds replaces it, and the file's other rules stay. It needs no PR
number. If the forge has no account of yours yet, log in to the forge once
first. If the classifier's account may not push to the repo, the page says so:
a repo admin adds it as a collaborator with write access
([the service account](review-classifier.md#the-service-account)).
#### Classifier settings
The **classifier settings** section shows the runtime, model and
effort the [review classifier](review-classifier.md) runs with and how many
requested reviews run at once, and lets you change them. The runtime is **API
key** (claude on the Anthropic key) or **ACP** (opencode on a provider). On the
API key runtime the model is free text, one word as claude's `--model` takes
it, such as `haiku` or `sonnet`. Choosing ACP shows the provider's fields in
place of the model: its `https://` URL, its model name, and what it charges in
USD per million tokens for input and output, with optional cache read and
cache write prices. The effort is one of low, medium, high, xhigh and max, or
the runtime's own default. Concurrent reviews is a whole number from 1 to 8;
try-outs don't count toward it.
**save** is on once something changed and the fields are valid. A save changes
the runtime, model, effort and limit, and the provider too while the runtime is
ACP, so switching back to API key keeps the stored provider for next time. If
the controller refuses it, the page shows the reason and leaves the form as you
typed it. A change takes effect from the next run, with no restart. The page
has no API key field: admins set each runtime's key in the bao UI.
### Scheduled prompts
`/schedules` has one card per scheduled prompt. The **name** is the card's
title, with the **agents** it fires for, the message, how often it repeats and
when it fires next. **edit** turns the card into its own form; **pause** and
**resume** stop and restart it; **remove** deletes it after a confirmation.
**+ schedule** opens an empty form above the list, and the filters narrow the
list by name, agent or state (`active`, `paused`, or `past` for a one-shot
whose time has passed). `past` is the controller's clock against the start time,
not a record that an agent fired, so you can still pause a `past` schedule.
A schedule repeats **every** some number of seconds, minutes, hours or days, or
fires **once**. Its start time is the moment a one-shot fires, and the origin of
the fire times of a repeating one. The form previews the next three fire times.
Changing the start or the repeat moves them; editing only the name, message or
agents doesn't, and the form warns when an edit would.
A start in the past fires the latest missed time once, right away, in every
agent chosen, and the form says so.
The form offers the agents of the roster that `/agents` lists. The page reloads
that roster with the schedules, so an agent created after you opened the page
shows up without a reload. If the roster fails to load, the form says so instead
of claiming there are no agents.
Only an operator can change a schedule here or with `swarmctl schedule`; an
agent has no way to add one. The page needs the controller's `/api/schedules`
routes.
### Notification flows
`/notifications` lists the [notification router's flows](notify-router.md#flows)
in two groups. Each **built-in** flow has an on/off switch; switching one off
stops the router running it from its next event, and switching it on again
restores it. Each **custom** flow shows its name and whether it's on. Every
flow draws as a graph, the one the editor shows but read-only: the forge events
it starts from on the left, the conditions after them, and the todo or review
request each branch ends in. A box cuts a long line of text with an ellipsis;
hover over the box to read all of it.
A built-in flow has a **duplicate to edit** button and a custom flow has **edit**
and **duplicate** buttons. **create flow**, at the top of the page beside the
refresh interval, starts a custom flow with just a source. Creating and
duplicating open the editor on a draft and make nothing: the flow exists once
you **save** it there. Leaving the draft without saving, or discarding it, leaves
the flows as they were.
A switch sends the whole flows document with the revision the page read, `0` if
it read no document. The controller refuses the switch when the stored revision
is another, including when someone wrote a document after the page read none, and
the page shows why. Reload to switch against the current document.
Each box in a graph says how many of the router's newest 50 forge events reached
it (**12 hits**, say), from a dry run of the saved flows (`GET /api/notify/dry-run`).
Below the flows, **recent deliveries** lists what the router delivered, newest
first: when, which flow and node, the event and issue, who got it, and whether it
went through (`GET /api/notify/deliveries`). Both come from the router; while the
running router is older than the controller, each shows the controller's "not
available" message in place of its numbers or table, and the rest of the page
works. See [deliveries and dry runs](notify-router.md#deliveries-and-dry-runs).
#### Editing a flow
`/notifications/<flow id>` edits one custom flow, and `/notifications/draft/new` makes
one: it opens on an empty flow, or on a copy of a flow with `?from=<flow id>`, as
**create flow** and **duplicate** do. A new flow takes its id from its name when
you save it, and **save** needs a name. The canvas draws the flow left
to right: the source on the left, each match to its right, and the leaf nodes at
the ends. Pick a node to edit it in the panel beside the canvas:
- a **source** picks the forge events the flow starts from;
- a **match** lists conditions, each an event field and the value it must equal;
- a **todo** says who gets it (the agents an event field names, or one agent, picked from the agents the swarm has),
whether it sets or clears a todo, the todo's key, and its text. The key and the
text can use `{field}` placeholders;
- a **request_reviewer** names the forge login the router asks to review the
pull request. It ends a branch, like a todo, and the row of kinds above the
canvas only offers it when the controller does.
Rewire the flow on the canvas:
- **Add** a node from the row of kinds above the canvas. Choose a kind to add it
below the picked box, or drag it onto any box.
- **Move** a node and its branch by dragging its box onto another box. With the
keyboard, the arrow keys walk the tree, **M** picks up the focused box, the
arrows go to the box you want it under, and **Enter** drops it there.
**Escape** puts it back. The panel's **move under** list does the same.
- **Reorder** the branches of a node. The router runs a node's branches in the
order they sit, so when a clear and an upsert share a key prefix, the order
decides what's left. A box with siblings says when it runs, such as "runs
2nd". **run earlier** and **run later** in the panel, or **Alt+Up** and
**Alt+Down** on the canvas, swap it with its neighbour.
- **Remove** a node and its branch with **remove** in the panel or the
**Delete** key.
The editor turns away any edit the controller would refuse to save, and says
why: a leaf node with children, a second parent, a node under its own branch, or
more nodes or levels than the controller allows. A drop it will refuse shows the
target box dashed and red while you hold the box over it.
Edits stay in a draft until **save**, which sends the whole flows document with
the revision the editor read. **discard edits** goes back to the saved flow and
**delete flow** removes it, after a confirmation. If someone changed the flows
elsewhere in the meantime, the controller turns the save down as stale, and the
editor offers to reload and drop your edits, or to keep them and replace this
flow's latest version with them. For any other refusal the editor shows the
controller's own message. The router runs a saved flow from its next event.
Where the controller's schema knows the values, the editor offers them instead of
asking you to type a name: `kind` and `event` conditions and the todo's `op` are
dropdowns, `action` and `repo` conditions and a reviewer are boxes that list the
known values to pick from and still take any other text (the repositories listed
are those with an open issue, so one with none is missing), and a todo's agent is a
dropdown of the swarm's agents. A value set in a flow that the list no longer
holds stays selected, marked "not listed." If a list fails to load, or the
controller predates the schema, the box is plain text. The values come from the
controller's schema ([notify-router](notify-router.md)), so a new one shows up
without a UI change.
**preview** runs the draft, saved or not, over the newest forge events and
shows on each box how many of them reached it, as a dry run does for the saved
flows in the list. It needs no save, so you can try a half-built flow. The
numbers go as soon as you edit the flow. If the draft would not save, or the
router can't run a dry run yet, the controller's own message shows instead.
Leaving the page with unsaved edits asks first, whether you close or reload the
tab, follow a link in the app, or press the browser's back button. Choose
**leave and drop edits** to go on, or cancel to stay and save.
### Sending a message to an agent
Under an agent's terminal, in the preview and in its own tab, is a **message**
box. **send** hands the text to the agent's harness, and the agent takes it as
an inbox message from your authelia username. **sent** means the harness was
listening and holds the message. If the harness isn't running, or isn't
connected to the swarm queue, the box shows an error and the agent gets nothing.
Subagents have no message box.
The message doesn't go through the agent's hive, so the hive dashboard's
message history doesn't list it. The agent's answer shows in its terminal; an
agent that calls `send` to your username gets an unknown-recipient error. A
harness that restarts before taking its turn loses the message.
<details><summary>How a message reaches the agent</summary>
The gateway sets `Remote-User` from authelia's answer on every `/api/`
request, replacing any the browser sent. `POST /api/agents/<name>/input`
takes that header as the sender and sends `{from, body}` as a NATS request on
`$SWARM.input.<name>`. swarm-controller is the only queue client allowed to
publish there. The agent's harness subscribes to its own subject and queues
each message for its turn loop with its reminders. Only the agent's own queue
credential may subscribe there. The harness doesn't reply, so swarm-controller
answers 202 when the request times out after two seconds, and 503 when the
queue reports no subscriber.
`POST /api/agents/broadcast` takes `{agents, body}` and sends the same
message to each agent the same way, up to 16 at once. It answers 200 with
`{results: [{agent, sent, reason?}]}`, one entry per agent in request order,
and `reason` holds the text the single-agent route would have answered with.
One agent not listening doesn't stop the others. swarm-controller refuses the
whole request and sends nothing when `body` or `agents` is empty, when
`agents` names an agent twice, names more than 64, or names one that
`GET /api/agents` doesn't list.
`GET /api/messages/stream` is an SSE feed of the messages between agents, and
from senders that are no agent (the operator, `reminder`, `compact`) to an
agent. It reads the terminal rows on `$SWARM.term.<agent>` and has no hive in
the path. Each event's `data` is `{ts, from, to, body, body_format?}`. An
agent's `send` row gives one event from that agent to the recipient in the
row's summary. A turn started by a non-agent sender gives one event to the
agent whose terminal holds the row. A turn started by an agent gives none,
since that agent's `send` row already did. The query `since_minutes` (default
60, at most 1440, 0 for live only) sends recent history first, oldest first,
read from each agent's newest 100 rows. A message an agent drained with an
in-turn `recv` starts no turn and so is missing.
The **messages** page (`/messages`) reads that feed. A **history** select picks
the window, from live only up to 24 hours, and a filter keeps the messages whose
sender or recipient contains the text you type. Each message is one row of the
same list an agent's terminal uses, a summary line with the time, the two
parties and the start of the body, and the whole body under it. The list
follows the newest message unless you scroll up, and holds the last 2000. The
feed has no event id, so a reconnect sends the history again, and the page
empties its list each time the connection opens, then fills it from the replay.
Changing the window opens a new stream.
A command has its own subject, so the harness never mistakes typed text for
one. `POST /api/agents/<name>/cancel` sends `"cancel"` as a NATS request on
`$SWARM.control.<name>`, and only swarm-controller may publish there. The
harness answers `{acted}`, which says whether a turn was running when the
cancel stopped it. It runs the same code as the hive dashboard's cancel button.
Unlike input, the harness answers a command, so swarm-controller returns that
answer as 200, 503 for no subscriber, and 504 when the harness held the command
and didn't answer within ten seconds. A 504 means unknown, not cancelled.
</details>
<details><summary>Where the subagent list and rows come from</summary>
The agent's subagent daemon publishes each subagent's terminal rows on
`$SWARM.term.<agent>.sub.<subagent>` with the agent's own queue credential,
into a stream named `term-sub-<agent>` that swarm-controller creates for every
declared agent. The stream keeps rows for 24 hours. swarm-controller serves the list as
`GET /api/agents/<name>/subagents`, the subagents named by the subjects in that
stream, each with `last_row_unix`, the stream's publish time of its newest row,
or `null` when that row isn't read within two seconds. The lookups run
concurrently, sixteen at a time, so the list waits at most two seconds for
each sixteen subagents.
swarm-controller also relays one subagent's rows as SSE on
`GET /api/agents/<name>/subagents/<subagent>/term/stream`. Like the agent's
terminal route below, it creates a short-lived consumer on that stream, here
filtered to the one subagent, which sends its newest 100 rows and then each new
one, with the same `Last-Event-ID` resume. The stream holds every subagent of
the agent, so for a fresh reader the consumer starts at the oldest row and
swarm-controller drops all but the newest 100 before sending. An agent with no
store identity, or no queue address, publishes no subagent terminals.
</details>
<details><summary>Where an agent's terminal history comes from</summary>
The harness publishes the agent's terminal rows on `$SWARM.term.<agent>`, into
a stream named `term-main-<agent>` that swarm-controller creates for every
declared agent, with the same 24 hours or 64 MiB as the subagent stream. That
stream holds this one subject, and the subagent stream holds only subjects
below `$SWARM.term.<agent>.sub`, so they never overlap.
`GET /api/agents/<name>/term/stream` creates a short-lived consumer on it
that starts 100 rows back, so the SSE stream sends those rows and then each
new one. Each event's `id` is the row's stream sequence, and a request with
`Last-Event-ID` resumes after it. An id past the newest row of the stream, as
after swarm-controller recreates the stream, starts over as a fresh reader. When the
consumer stops sending its heartbeat, as after a queue restart, the route ends
the response and the browser reconnects. Until the stream exists, the route sends
live rows only.
</details>
### Issue report
The **issues** page lists the open issues of every repo that has one, from
`GET /api/issue-report`, or of one repo from
`GET /api/repos/{org}/{repo}/issue-report`. Each row shows whether an open
dependency blocks the issue and how many other issues depend on it. The
list sorts by that count by default, and **hide blocked** drops the blocked
rows.
Picking a repo fetches its report again. The page remembers the pick, so a
repo that no longer has an open issue stays in the **repo** select with the label
**no open issues**, and the table stays filtered to it. The repo list and the report
fail on their own: a repo list that fails to load shows its error beside the
report, and a report that fails to load clears the table instead of leaving the
previous repo's issues under the error.
### Builds
The **builds** page lists the agent config PR heads the swarm builder built,
newest first, from `GET /api/builder/runs`. Each card shows the repo and PR,
the first seven characters of the commit, how the build ended (**built**,
**failed** or **error**), when it started and how long it took. Select a card to
read it: for a failed build, what `nix` printed; for a built one, the system
closure; for an error, why no build came back.
The **Details** link on a `swarm-builder` commit status opens this page as
`/builds?run=<id>`, with that build's card open and in view. The builder keeps
its newest 100 builds, so a link to an older one shows a note instead.
### Broadcasting a message
The messages page (`/messages`) has a **broadcast** card above the message list. It
starts folded; select its header to open it. The card holds a composer that sends one message
to several agents at once, and the messages show up in the list below as they
go. Folding it again keeps what you typed and ticked. Pick the
agents from a grid of their icons. **running** shows the agents that report
themselves running and **all** shows every agent; the browser keeps the choice.
**select all** ticks every agent the filter shows, and **clear** unticks them.
The composer skips an agent you ticked and then hid with the filter. Type the
message and press **send to N agents**. The text goes exactly as typed, as an
inbox message from your authelia username, the same as the box under an agent's
terminal. One broadcast may name at most 64 agents, and the button stays off
until you untick the extra ones. A broadcast to many agents can take a few
seconds to answer, and the button reads **sending…** meanwhile.
The composer then lists what happened to each agent. An agent that wasn't
listening shows as not sent, with the reason, and the others still got the
message. **select the N that failed** ticks just those, switching to **all** if the
filter hid any, so you can send the same message again. When every agent got the message the box clears; otherwise the
text stays.
### Inference settings
The agent's detail panel shows its model and effort under **inference**, each
with where it comes from: the swarm, its agent.nix, its own pick, or the
default. Below them are the subagent model list the swarm sets, or "its own"
when it sets none, and the subagent cap, which reads "no cap" when unset.
**edit** opens a dialog with the four fields, filled with what the swarm
stores; saving does a `PUT /api/agents/{agent}/inference` of
the whole value. A field you leave empty stays out of it, so the agent keeps
its own value for that field; an empty max subagents is no cap. **clear the swarm's setting** sends the `DELETE`.
The panel reads `GET /api/agents/{agent}/inference`; before the agent's first
report reaches the controller, it shows the swarm's stored values only.
[inference](inference.md) describes the settings themselves.
### Editing an agent's instructions
The **instructions** row of an agent's detail panel has an **edit** button. It
opens `AGENTS.md`, the file the agent reads its instructions from, in a plain
text box. The text is the copy the agent last published, so the box still
opens for a stopped agent, but **save** needs the agent running.
Save sends the swarm a diff against the version you opened, and the agent
applies it to its own file with git. When the agent changed the file in the
meantime, the swarm refuses the diff. The box then keeps your text, shows the
agent's version below it, and offers two ways on: **keep my text, replace the
agent's changes**, which makes your text the new base for the next save, or
**use the agent's version**, which drops yours. A diff that git refuses for any
other reason shows git's message and leaves the file as it was. Closing the
dialog keeps an unsaved edit for the next time you open it.
### Resource caps
The agent's detail panel shows its CPU quota and memory max under **resource
caps**. A cap is the swarm's, so it follows the agent to whichever hive runs
it, and it replaces what the hive says for that field. A field the swarm
leaves out shows as the hive deciding it. **edit** opens a dialog with the two
fields, filled with what the swarm stores. A cpu quota is a percentage of one
core (`400%` is four cores). A memory max is a size such as `512M` or `8G`, a
percentage of the machine's memory, or `infinity`.
Saving does a `PUT /api/agents/{agent}/limits` of the filled fields. A `PUT`
replaces the stored value whole, so an empty box clears that cap and the hive
decides that field again. **clear the swarm's caps** sends the `DELETE`. The
controller checks each value against systemd's own syntax, because systemd
ignores a value it can't parse and the agent would run uncapped, and it
refuses, for example, a percentage with a leading zero. The dialog shows the
refusal's reason under the fields and leaves them as you typed them.
A saved change isn't instant. The hive reads the caps about every 30 seconds,
and a running agent takes a changed cap at its next spawn or rebuild, so the
row shows the new value before the agent runs under it. The panel reads
`GET /api/agents/{agent}/limits`; when that read fails, the panel shows the
reason and a **retry** button. [hivectl](../tools/hivectl.md#per-agent-resource-limits)
sets the hive's own limits.
### Permissions
The agent's detail panel shows what the swarm grants it under **permissions**:
the default set, none, or a count with one badge per permission. A swarm
grant is the agent's whole set, so it follows the agent to whichever hive runs
it. **edit** opens a dialog with one checkbox per permission the controller
knows (`GET /api/permissions`), ticked as the grant stands, or on the default
set when the swarm sets none.
Saving does a `PUT /api/agents/{agent}/permissions` of the ticked boxes. A
`PUT` replaces the stored list whole, so an unticked box revokes that
permission, and ticking nothing grants nothing. Ticking exactly the default set sends the
`DELETE` instead, as the create form sends no list for it, so the agent follows
the default if it changes. **use the default** sends the `DELETE` too. The
dialog shows a refusal's reason under the boxes and leaves them as you ticked
them.
A saved change isn't instant. The hive reads the grant about every 30 seconds
and the agent takes it at its next turn. A change to `manage_root_agent`
restarts the agent after its current turn. The panel reads
`GET /api/agents/{agent}/permissions`; when that read fails, the panel shows
the reason and a **retry** button. `swarmctl agent permissions set` sets the
same grant from a shell.
### Creating an agent
The **+ agent** button opens a dialog of five steps, which you go through in
order with **next** and can step back through with **back** or the step tabs:
1. **identity**: the name, the hive, and optionally an
[archetype](archetypes.md) that fills in the settings of the steps after it.
**next** waits for a valid name (1-63 lowercase letters, digits and hyphens)
and a hive.
2. **instructions**: the text the agent's `AGENTS.md` starts with, in a box
of its own.
3. **permissions**: one checkbox per permission, starting from the default set
or the archetype's.
4. **resources and inference**: the optional resource caps and inference
settings.
5. **summary**: every setting the request will carry, and what stays at its
default, then **what this creates**. **create** is on this step only.
The header, the step tabs and the buttons stay in place and only the step's own
fields scroll, so a long step never pushes **next** out of view. After
**create**, the dialog says it queued the job and links to **jobs**, and offers
**create another** and **close**. **create another** starts on step 1 with every field back at
its start, and keeps only the hive you picked. If the controller refuses the `AGENTS.md` text,
the dialog goes back to step 2 and shows the reason there.
### Settings at creation
Step 4 of the **+ agent** dialog has **resource caps** and **inference**, with
the same fields as the detail panel's editors. Leave the
fields empty and the request is the plain name and hive. Fill them and `POST /api/agents` carries them
as optional `limits` and `inference` objects, the same shapes the two `PUT`
routes take. The controller checks them with the `PUT` rules before it queues
anything, so a bad value gets a refusal with its reason and creates no agent.
The dialog loads the hive list and the permission list when it opens. If either
load fails, the dialog shows the reason in an error panel above the step's fields, and
the hive picker or the permissions area reads `could not load hives` or
`could not load permissions` in place of `loading…`.
It stores them ahead of the job graph, before it declares the agent, so they
are in place before the agent first runs. If the creation then fails, the
controller puts back what it replaced. Each agent name keys its settings and
destroying an agent doesn't clear them, so a name set up before keeps its
stored settings when you leave the fields empty.
### Initial AGENTS.md at creation
Step 2, **instructions**, holds a plain text box for the
text the new agent's `AGENTS.md` starts with. Left empty, or holding only
whitespace, it sends nothing and the agent starts without the file. Otherwise
the text goes as typed in the optional `agents_md` field of `POST /api/agents`.
The agent writes it on its first start, and only when it has no `AGENTS.md`
yet, so creating an agent again under a name that has one never replaces it.
The controller refuses a text over 64 KiB. Its reason shows under the
box, and the swarm creates no agent. After a successful
create the box empties. The
[instructions editor](#editing-an-agents-instructions) changes the file later.
The [swarm README](README.md#an-agents-agentsmd) has the wire details.
### Linking external accounts
Each agent on `/agents` opens three dialogs that write a credential for it
into the swarm secret store through swarm-controller. All three are blind
set actions: no route hands a token back. swarm-controller refuses to link a
name that already holds an account; delete that account first to replace it.
The agent's detail panel lists its linked accounts under **accounts**, one row
per account: kind, name and host. Opening an agent makes one request,
`GET /api/hives/{hive}/agents/{agent}/linked-accounts`, which returns every
account of that agent as names and hosts and never a credential. When the panel
fails to read the accounts, the inference settings, the resource caps or the
permissions, it
shows the reason as text. A list the panel already loaded stays on screen
beneath the error.
| kind | one row per | name | host |
| ------- | ------------------------------------------- | ----------- | ------------------------ |
| matrix | `swarm/agents/<agent>/matrix/<account>` | the account | its `homeserver`, if set |
| forgejo | `swarm/agents/<agent>/forge/<label>` | the label | its `url` |
| github | `swarm/agents/<agent>/github-token`, if any | `github` | `github.com` |
The route names the second kind `forgejo`; the panel shows it as **forge**,
the word the link menu uses.
The matrix `main` row is the agent's own account, which the swarm mints;
it carries an **own account** badge. The panel requests the list again when
a link dialog closes.
- **link a matrix account** — `PUT /api/hives/{hive}/agents/{agent}/matrix-accounts/{account}`,
either a pasted bearer token or a user id + password that
swarm-controller logs in with, storing the token it gets back.
- **link a forge account** — `PUT /api/hives/{hive}/agents/{agent}/forge-accounts/{label}`
with a forge URL and an access token, stored at
`swarm/agents/<agent>/forge/<label>`. The agent's
`hive-agent-forge-accounts` unit fetches it into
`<state>/forge-<label>-token` and `<state>/forge-<label>.json`, the files
`hive-forge -f <label>` reads. The unit rewrites a pair whenever the
store's account for its label differs, and never deletes one: a pair whose
label the store doesn't list stays untouched. swarm-controller answers 409
to a link for a label that already holds an account, so replacing one
takes a delete, then a new link.
- **link a github account** — `PUT /api/hives/{hive}/agents/{agent}/github-account`
with a personal access token, stored at `swarm/agents/<agent>/github-token`.
One token per agent. The agent's
`hive-agent-github-token` unit fetches it into `<state>/github-token`,
the file its `gh` wrapper, git credential helper and GitHub notification
poller read. A `github-token` already in place stays when the store holds
none. What the token needs and how the agent uses it:
[GitHub accounts](../integrations/github.md).
#### Deleting a linked account
Every row except the matrix `main` row has a **delete** action. Its
confirmation names the account and its host, and confirming sends one
request:
| kind | route |
| ------- | ------------------------------------------------------------------- |
| matrix | `DELETE /api/hives/{hive}/agents/{agent}/matrix-accounts/{account}` |
| forgejo | `DELETE /api/hives/{hive}/agents/{agent}/forge-accounts/{label}` |
| github | `DELETE /api/hives/{hive}/agents/{agent}/github-account` |
swarm-controller removes every version of the entry from the swarm secret
store, and answers 404 when the store holds nothing there. It refuses `main`: the
swarm mints that account and would re-mint it. The panel requests the list
again after a delete, including a 404 one: the account being already gone is
the end state the delete wanted, so the dialog closes without an error.
The matrix confirmation has a checkbox, off by default, that logs the token
out at its homeserver first (`?revoke=true`). A 401 `M_UNKNOWN_TOKEN` from
that logout means the homeserver already doesn't recognise the token, so the
delete proceeds as if it had succeeded. Any other failed logout, or an account
with no homeserver stored, leaves the account in the store and the dialog
shows why. Deleting a forge account or GitHub token leaves the token valid at
its provider; revoke it there.
The agent isn't told about a delete. Its matrix daemon drops the account when
it next lists the store, within two minutes. Its forge and GitHub units never
delete a file, so a token already fetched into `<state>` stays there.
Where each credential lives and who reads it:
[`credentials.md`](credentials.md).
## Enabling
```nix
services.hyperhive.deploy.swarm-controller.enable = true; # the UI comes with it
```
The UI has no switch of its own (setting `deploy.swarm-ui.enable` or
`swarm.ui.enable` has no effect, and warns at evaluation), and doesn't follow `allSwarmServices`:
the UI is a view onto the controller's state and reaches it over that
daemon's socket, so the host that runs the controller is the host that
can serve the UI. A hive that merely _uses_ a swarm has nothing to serve.
The UI build is `deploy.swarm-controller.ui.package` (defaulted from this
flake's `swarm-ui` package). The older `deploy.swarm-ui.package` still works
and warns at evaluation.
The UI answers on `services.hyperhive.swarm.domain` and nothing else. It
shares that name with the swarm-controller it fronts — one service to a
reader and to a certificate — so there is no separate option to pin.
> The apex must differ from `services.hyperhive.domain`. The gateway's
> default server already answers for the hive domain, and two vhosts
> claiming one `server_name` don't error — nginx picks one — so this is
> an assertion rather than a runtime surprise.
## 🔑 You must be in the `admins` group
**This is the step that separates "protected" from "locked out."** The
vhost's `auth_request` asks authelia _"is there a session"_; the rule
that makes it mean _"is this an operator"_ is an `access_control` entry
requiring `group:admins`, followed by a `deny` for everyone else. An
account without that group authenticates fine and still gets bounced.
```sh
swarmctl user add <you> --email <you>@example.com --group admins
swarmctl user update <you> --add-group admins # an account that already exists
```
`--email` isn't needed for the UI, but the forge won't create your account
without one → [setup.md § 2](../getting-started/setup.md#2--your-sso-account).
Why a group and not "any session": agents are authelia subjects too, so
_authenticated_ includes every agent in the swarm. The group is the only
thing standing between "an operator's page" and "anyone with a session."
## What it costs to be reachable
hyperhive publishes the apex to the hive's resolver like every other
swarm service, so agent containers can resolve it. That's deliberate and it's
not a hole: **reachability isn't the access control here.** An agent
that resolves the name and connects still has no operator session, and
the subrequest denies it.
## Quick links
The swarm UI's header carries a single 🔗 button, visible on every route,
opening a popover of links to other swarm-wide services. Backed by
`GET /api/links` (swarm-controller), which serves
`services.hyperhive.swarm.controller.links` (a `listOf { label, icon, url }`,
same shape as the per-agent `services.hyperhive.agent.dashboardLinks`).
Links to swarm services come from swarm-level options, so the list is the
same whichever host runs each service:
| link | URL built from |
| -------- | ------------------------------------------------- |
| Forge | `services.hyperhive.swarm.forge.domain` |
| Authelia | `services.hyperhive.swarm.authelia.domain` |
| Grafana | `services.hyperhive.swarm.grafana.domain` |
| Metrics | `services.hyperhive.swarm.victoriametrics.domain` |
| Logs | `services.hyperhive.swarm.victorialogs.domain` |
| Matrix | `services.hyperhive.swarm.matrix.gatewayHost` |
| Bao | `services.hyperhive.swarm.bao.ui.domain` |
Each entry is present whenever its option has a value — every option in
the table defaults to one, so the link appears even on a swarm that runs
no instance of that service.
`nix/host-modules/swarm-controller.nix` builds these entries. `swarm-ui.nix`
adds one more, for this UI's own API docs. An operator can add entries
directly. An empty list hides the button.
The **Matrix** entry opens the swarm's matrix web client (fluffychat,
`services.hyperhive.deploy.matrix.gui.package`) at the homeserver's
gateway host, `chat.<swarm domain>` by default. The homeserver's host
always serves the client at `/` on that vhost, so the entry is present
whenever the swarm names a gateway host.
The **Bao** entry opens the secret store's web UI, which admits members of
authelia's `admins` group only; see [`secrets.md`](secrets.md#the-browser-ui).
<details><summary>Adding a swarm service name: the two wiring sites</summary>
Adding a swarm service name means touching two things. Missing the
second ships as a different flavour of "works from the host, broken from
a container":
| site | file |
| ---------------------------- | ---------------------------------------------------------------------- |
| vhost + `gateway.localNames` | the service's own module (for example `nix/host-modules/swarm-ui.nix`) |
| **certificate name** | `nix/host-modules/swarm.nix` (`serviceDomains`) |
<!-- vale write-good.Passive = NO -->
The DNS record and the local-dev `/etc/hosts` entry need no separate
edit: both derive from `services.hyperhive.gateway.localNames`,
which a service's own module already has to push its domain into to be
resolvable — see `nix/host-modules/hive-gateway/dnsmasq.nix` and
`.../default.nix`'s `networking.hosts`. `vhosts.nix` itself is scoped to
the surface the hive's own domain serves (dashboard, per-agent routing,
pinned-hive discovery); a swarm service declares its own vhost next to its
own options, the way `swarm-ui.nix` and `swarm-authelia.nix` do.
<!-- vale write-good.Passive = YES -->
⚠️ The certificate one is the hardest to predict and the most visible when
missed. `serviceDomains` is _both_ the `allowed_domains` the secret
store's `pki/roles/swarm-services` narrows to and the list each gateway's
leaf draws its SANs from (it carries the ones that host fronts),
and the apex is a **sibling** of `forge.<swarm>` / `chat.<swarm>` /
`auth.<swarm>`, not a parent — no CA in the hierarchy issues for it
implicitly. Left out, the vhost falls back to the hive leaf and the
swarm's front page opens with a name mismatch.
</details>
## Cross-references
- [`sso.md`](sso.md) — the authelia instance itself, and the user store.
- [`../networking/gateway.md`](../networking/gateway.md) — the full vhost map and TLS modes.
View git blame Copy permalink