Watch
0
0
Fork
You've already forked hyperhive
0

swarm ui: agents page opens an agent's details beside the list

Selecting an agent card opens its details in a column on the right of the
agents list, at /agents/<name>, instead of unfolding the card in place. The
list is full width until an agent is selected, then narrows and the details
slide in. A narrow screen shows the details alone, with a back button. A
card's message is cut to one line.

The shell treats /agents/<name> as the agents page, so the nav tab stays lit
and the page is not rebuilt on each selection.
This commit is contained in:
iris 2026-11-22 18:42:20 +01:00
commit 7bdaea7954
2 changed files with 494 additions and 154 deletions

View file

@ -13,7 +13,7 @@ place, and every one reads from [`swarm-controller`](../../../swarm-controller/R
| route | what |
| ---------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `/` | the start page: the swarm's name over a digital-rain backdrop with the slug (`hyperhive / we are the wired`, its shaded gradients pulsing in a wave unless reduced motion is on) at the foot, plus a paperclip offering to set up an agent while no `ruth` exists, or her hive declares her destroyed (it opens the create form with `ruth` filled in and the `ruth` archetype picked); the swarm name in the header links here |
| `/agents` | every agent as a card that opens in place (status, terminal, subagents, and settings with permissions, internet, allowed domains and accounts); create agents (with their permissions, optionally from an [archetype](archetypes.md)), link forge, matrix and GitHub accounts |
| `/agents` | every agent as a card; select one to open its details beside the list (status, terminal, subagents, and settings with permissions, internet, allowed domains and accounts); create agents (with their permissions, optionally from an [archetype](archetypes.md)), link forge, matrix and GitHub accounts |
| `/swarm` | what belongs to the whole swarm: a folded **hives** card with the hive directory and each hive's last reported status, a folded **egress** card with the swarm-wide allowed and denied domains, and a folded, **beta** **notification router** card (with a warning banner) holding the router's flows: a switch for each built-in, a card for each flow an archetype carries, every flow drawn as a graph, and an editor for custom flows. `/notifications` redirects here |
| `/agents/<name>/terminal` | one agent's live terminal |
| `/agents/<name>/subagents/<subagent>/terminal` | one of that agent's subagents' live terminal |

View file

@ -62,7 +62,7 @@ asks from inside a handler with the `useConfirm` hook, with no dialog state of i
| route | what |
| ---------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `/` | the start page: the swarm's name over a digital-rain backdrop with the slug (`hyperhive / we are the wired`, its shaded gradients pulsing in a wave unless reduced motion is on) at the foot, plus a paperclip offering to set up an agent while no `ruth` exists, or her hive declares her destroyed (it opens the create form with `ruth` filled in); the swarm name in the header links here |
| `/agents` | every agent as a card that opens in place (status, terminal, subagents, and settings with permissions, internet, allowed domains and accounts); create agents (with their permissions, optionally from an [archetype](../../2-handbook/swarm/archetypes.md)), link forge, matrix and GitHub accounts |
| `/agents` | every agent as a card; select one to open its details beside the list (status, terminal, subagents, and settings with permissions, internet, allowed domains and accounts); create agents (with their permissions, optionally from an [archetype](../../2-handbook/swarm/archetypes.md)), link forge, matrix and GitHub accounts |
| `/swarm` | what belongs to the whole swarm: a folded **hives** card with the hive directory and each hive's last reported status, a folded **egress** card with the swarm-wide allowed and denied domains, and a folded, **beta** **notification router** card (with a warning banner) holding the router's flows: a switch for each built-in, every flow drawn as a graph, and an editor for custom flows. `/notifications` redirects here |
| `/agents/<name>/terminal` | one agent's live terminal |
| `/agents/<name>/subagents/<subagent>/terminal` | one of that agent's subagents' live terminal |
@ -86,7 +86,7 @@ The pages that only display (agents, alerts, builds, review runs and the hives
card) show the answer they last saw at once, then replace it with the fresh one
while the spinner runs, so opening one never starts on an empty page. On the very
first visit, the hives card and the alerts table show their own header with three
grey placeholder rows until the first answer comes. In an opened agent card, a
grey placeholder rows until the first answer comes. In an agent's details, a
setting still loading (permissions, resource caps, internet, allowed
domains) shows a grey pill the size of its badge, and the AGENTS.md editor a grey
block. The browser
@ -110,7 +110,7 @@ the code back. A failed login offers **try again** in the same dialog. When the
secret store has the login, the dialog closes and a
short toast in the top-right corner says so. The credentials go to the agent through the secret store
([`claude-login.md`](../../2-handbook/swarm/claude-login.md)). The agent's own web UI login still
works too. The settings section of an opened card has a **claude login** row: when
works too. The settings section of an agent's details has a **claude login** row: when
the swarm stored the agent's login, or that none is, with a **log in** button
(**log in again** when the swarm holds one) that opens the same dialog, and a **delete
login** action that asks first.
@ -130,21 +130,32 @@ sweep. A hive that doesn't publish the field shows no badge.
`/agents` is a column of separate cards: a filter card, then one
card per agent, none of them inside a bigger card. A card's
header shows its icon, name, the status message with how long ago the agent set it (`5m ago`), a **terminal**
header shows its icon, name, the status message on one line (the whole message is in the details and in the card's tooltip) with how long ago the agent set it (`5m ago`), a **terminal**
button that opens the agent's terminal tab, the **wanted** menu (up, offline,
paused, destroy), a **paused by hive** badge when the agent's own harness paused
its turn loop (hover for the reason; set the agent **up** to resume it), and a
**login needed** badge with a **log in** button when its claude has no session. Select the header to open the
card in place; select it again to close it. Several cards can be open at once,
and a card stays open across the page's refresh, but not across a reload. The
**login needed** badge with a **log in** button when its claude has no session.
Select a card to open the agent's details. The address becomes
`/agents/<name>`, so a link or a reload lands on the same agent, and the
browser's back and forward buttons walk through the selections. Until you select
an agent the list takes the whole width. Selecting one narrows the list to a
third of the page and slides the details in on the right, which scroll on their
own while the list stays in view, and the selected card gets a purple border and fill. The widths animate,
and stand still where motion is off. A screen narrower than 60em has room for one
column, so the details replace the list. At any width, the **← agents** button above
the details shuts them and the list takes the width again, and focus returns to
the agent's card. A name the roster doesn't have, from a stale link, shows
"no agent named …" and the same button. The
filter card, which also holds the **+ agent** button and the refresh picker,
filters by name, hive, status, message, wanted state
and config PR. While the agent list loads for the first time, the page shows grey
placeholder cards with the same layout as the real ones, so the cards don't jump
when the data arrives. The placeholders stand still where motion is off.
An opened card has six sections, and each reads nothing until you open it.
**status** and **terminal** start open, the other four folded:
Below the agent's state, message and wanted menu, the details have six sections,
and each reads nothing until you open it. **status** and **terminal** start
open, the other four folded:
- **status**: how fresh the agent's last report is (with its age), its turn state and context, its hive, links to its
profile and its config repo on the swarm's forge (built from the forge entry
@ -169,7 +180,7 @@ An opened card has six sections, and each reads nothing until you open it.
### Agent and subagent terminals
An opened agent card on `/agents` shows, in its **terminal** section, a small live preview of its
An agent's detail pane on `/agents` shows, in its **terminal** section, 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
@ -273,7 +284,7 @@ with no icon keeps the hyperhive mark. Everywhere else the swarm UI shows an age
The tab also has an agent badge, with the agent's icon (carrying the connection
light) and name, ahead of the state badges. It opens the agent's details in a popover under the badge:
the same status, state, wanted state and folded **status** and **settings** sections (inference, resource caps, internet, allowed domains, permissions and linked accounts) as the
opened card on `/agents`, without the terminal and subagents. The popover doesn't dim the
details on `/agents`, without the terminal and subagents. 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
@ -704,7 +715,7 @@ store identity, or no queue address, publishes no subagent terminals.
<details><summary>Where the background task list comes from</summary>
An opened agent card has a **background tasks** section, below the subagents, listing the
An agent's detail pane has a **background tasks** section, below the subagents, listing the
tasks the agent started through the bash MCP. Its bash MCP publishes one status per task
to the `bash-task-status` bucket, and swarm-controller serves the list as
`GET /api/agents/<name>/bash-tasks`: each task's id, command, state, the time the agent queued it and
@ -832,7 +843,7 @@ text stays.
### Inference settings
An opened agent card shows, in its **settings** section, its model and effort under **inference**, each
An agent's detail pane shows, in its **settings** section, 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, the subagent cap, which reads "no cap" when unset, and the
@ -847,7 +858,7 @@ report reaches the controller, it shows the swarm's stored values only.
### Editing an agent's instructions
The **instructions** row of an opened agent card's **settings** has an **edit** button. It
The **instructions** row of an agent's **settings** 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.
@ -863,7 +874,7 @@ dialog keeps an unsaved edit for the next time you open it.
### Resource caps
An opened agent card shows, under **settings**, its CPU quota and memory max under **resource
An agent's detail pane shows, under **settings**, 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
@ -904,7 +915,7 @@ reason and a **retry** button.
### Permissions
An opened agent card shows, under **settings**, what the swarm grants it under **permissions**:
An agent's detail pane shows, under **settings**, what the swarm grants it under **permissions**:
a badge reading the default set, none, or a count. A swarm
grant is the agent's whole set, so it follows the agent to whichever hive runs
it. Selecting the badge opens a box under it with one checkbox per permission the controller
@ -975,7 +986,7 @@ same grant from a shell.
### Allowed domains
An opened agent card lists, under **settings**, the destinations it may reach through its
An agent's detail pane lists, under **settings**, the destinations it may reach through its
hive's egress proxy under **allowed domains**, beyond the base list every
agent has: one badge per source with its count, **swarm-wide** (the entries for every
agent), **this agent** (the ones for it alone) and **automatic**. Selecting a badge
@ -1057,7 +1068,7 @@ 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 opened card's settings editors. Leave the
the same fields as the agent's settings 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
@ -1100,14 +1111,14 @@ reads back and its box starts with the stored text. The secrets never read back,
so their box starts empty and replaces the stored secrets only when filled and
saved.
The **claude login** row of an opened agent also sets the API key: **set API key**
The **claude login** row of an agent's details also sets the API key: **set API key**
(or **replace API key**) sends the key as `ANTHROPIC_API_KEY=<key>` to the
agent's secrets, **remove API key** deletes them, and **advanced** opens
the form above. The row shows whether the store holds a key and never the key. A
change reaches the agent at its next restart, and an agent with a key tries it
instead of waiting for a claude login.
An opened agent card lists its linked accounts at the end of its **settings** section, one row
An agent's detail pane lists its linked accounts at the end of its **settings** section, 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 card