The agent detail panel shows the model and effort an agent runs on with
where each comes from, plus the swarm's subagent model list. An edit
dialog sets the swarm value (PUT) and a clear action removes it (DELETE).
Reads the configured/effective view, and shows the stored values alone
until an agent has reported what it runs on.
The header on `$SWARM.agent-state.<agent>` went out only when a bus event
changed it, and the core subject keeps nothing for a late subscriber. An
idle agent therefore said nothing to a reader that attached after its
last change: swarm-controller after a restart showed `effective: null`
for that agent until its next turn, which can be hours.
The publish loop now also wakes every 60 s and re-sends the current
header through the same `publish` path (clearing the dedup state first).
Transitions still go out as they happen. Same subject, same grant.
An agent whose queue connection is down now logs `swarm agent state:
publish skipped` once a minute even while idle, so the
`hh-swarm-agent-state-publish-skipped` alert also fires for idle
disconnected agents, not only busy ones.
Refs #5034
Each agent now reports the model and effort it runs on, and where each
comes from (`swarm`, `agent_nix`, `local_file` or `default`), in its
turn-state header on `$SWARM.agent-state.<agent>`: `effort`,
`model_source` and `effort_source` join `model`. The shared
`swarm_queue_client::agent_inference::{Effective, Source}` types spell
those fields for both ends.
The Bus keeps a source next to the model and effort. `Bus::new` takes it
from the local fallback chain (`harness_state::local_model` /
`local_effort`, which now return the source with the value), a `/model`
or `/effort` pick sets `local_file`, and the swarm poller sets `swarm`
or, on a clear, the local chain's source. A swarm value equal to the one
the agent already runs is now applied once, so the source moves to
`swarm` and the header is republished; before, `next()` skipped it as
unchanged.
swarm-controller subscribes to every agent's header with the reader
grant it already holds for the state relay, and keeps the latest
inference fields per agent in memory (`inference::Reported`).
`GET /api/agents/{name}/inference` now answers
`{configured, effective}`, `effective` being `null` until the agent's
first header with these fields since the controller started.
`swarmctl agent inference get` prints both; `set` reads `configured`.
Refs #5034
A start in the past fires the latest missed time once, right away, so the
form says that instead of 'already passed'. A one-shot past its time reads
'past' / 'time passed' (the controller's clock, not a record of firing) and
keeps its pause button. The edit form compares against what it opened with,
so a schedule changing under it is not reverted on save.
/schedules lists the controller's scheduled prompts as cards: name as the
title, the agents it fires for, the message, its cadence and next fire.
Edit turns a card into its own form; pause, resume and remove act in place,
and + schedule opens an empty form. The form sends only what changed, so a
rename never moves the fire times, and warns when the start or the repeat
does. Filters reuse the shared card filter bar.
Adds a TextAreaField to the ui kit, and an upcoming mode on RelativeTime so
a next-fire time reads "in 5m". Needs the controller's /api/schedules
routes.
All badges in the floating row take the interactive 2.75em height so the
state badges match live and expand. The embedded preview wraps them onto a
second row in the narrow details panel, so its top padding now fits two
rows (and the preview cap grows to match); the full tab keeps one row.
The floating row over the terminal now spans its top edge: the state
badges (connection first, then the turn-state ones, which move over from
the top-right corner) on the left, the expand trigger on the right.
The connection badge takes the trigger's height. The row passes pointer
events through its empty middle.
They sat in a toolbar row above the terminal box. They now float in
its top-left corner, next to the turn-state badges in the top-right,
and the toolbar row is gone. The full-tab mount keeps the connection
badge and, being the destination already, has no expand trigger.
The subagent list endpoint now returns {name, last_row_unix} per
subagent, the publish time of its newest terminal row, read from the
queue controller-side so agents publish nothing new. The panel sorts
newest first, shows a ticking age beside each name, and dims one quiet
for 15 minutes.
Closes#4928
The three link-account badges were wider than the agent detail panel and
spilled out of it; they are now a single badge opening a dropdown of the
three kinds. The subagents row, heading included, is not rendered for an
agent that has none.
The agent terminal tab now takes the whole window except the header bar,
full width and height, instead of sitting in the padded content column.
The connection badge floats inside the terminal's top-left corner and the
toolbar row above it is gone on this route.
Codeberg does not accept AI-assisted contributions, so it should not be
the example we put in front of operators. The placeholders now read
myforge / forge.example.org; the tests use the same neutral names.
The agent view in the swarm UI gets a message box. The text reaches the
agent's harness straight from swarm-controller over the swarm queue and
becomes an inbox turn from the operator's authelia username. hive-c0re is
not in the path.
- gateway: `/api/` forwards authelia's `Remote-User`, replacing any the
client sent.
- swarm-controller: `POST /api/agents/{name}/input` takes that header as
the sender (401 without it) and sends `{from, body}` as a NATS request
on `$SWARM.input.<agent>`. No subscriber is a 503.
- swarm-nats-auth: the controller is granted publish on `$SWARM.input.*`;
no hive or agent grant names the family, so the harness can trust
`from`.
- hive-agent: a task subscribes to the agent's own input subject and
pushes each message into the serve loop's existing reminder channel,
renamed `local_msg_*` since it now carries both.
The harness does not reply to the request: no agent grant covers a reply
subject. The controller therefore answers 202 once the request times out
(2s) and 503 on the server's no-responders status.
Closes#4828
The matrix, forge and github link routes wrote their credential
unconditionally, so linking a name that was already linked replaced the
working account. For matrix that lost the device the agent's crypto store
belongs to (#4838).
Each route now reads the account's store path first and answers 409,
naming the existing account, when something is stored there. Nothing is
written. Replacing an account takes the delete from #4899, then a link.
The matrix route checks before password mode's login, so a refused link
mints no new device at the homeserver.
The check is a read then a write, not an atomic step; two concurrent
links to one name can still both pass it.
Closes#4856
An agent's subagent daemon publishes each subagent's output as terminal
rows on `$SWARM.term.<agent>.sub.<subagent>`, as the agent, into a
per-agent stream it creates itself; swarm-controller lists an agent's
subagents from that stream's subjects and relays one subagent's rows as
SSE; the swarm UI lists them under the agent's terminal preview and
reuses AgentTermPreview, full-screen tab included, with no input.
- swarm-nats.nix: the agent token may also publish
`$SWARM.term.{agent}.sub.>` and `$JS.API.STREAM.CREATE|INFO` on
`term-sub-{agent}`, and nothing else of JetStream. A module-eval arm
pins the agent-token grant as an exact list.
- mcp.nix: hive-subagent-daemon loads the agent's store identity
(`hive-agent-bao-cert/-key/-server-ca`, the ones hive-agent loads)
whenever the agent has a store, not only on the opencode preset. The
agent's own queue secret lives in the store, so this is the credential
the harness connects with.
- hive-subagent-mcp: `swarm_term` reads the agent's queue secret under
that identity, connects with the agent token, opens or creates
`term-sub-<agent>` (max_age 24h), and publishes classified rows from
the sink every subagent line already passes through. The sink only
queues (bounded, drop-and-count); a missing store, refused credential,
failed stream create or failed publish is a log line.
- The stream-json classifier (`stream_enrich`) and the `TermMsg` row
types plus `fit` move from the hive-agent binary into hive-sh4re, so
the subagent daemon publishes the rows AgentTermPreview already
renders. hive-agent keeps its LiveEvent classifier on top.
- swarm-controller: `GET /api/agents/{name}/subagents` and
`GET /api/agents/{name}/subagents/{subagent}/term/stream`.
- docs/swarm: what the UI shows and what the queue carries.
Closes#4827
Addresses argus review comment 90297 on PR #4899:
- swarm-controller/README.md: list the three DELETE routes (including
matrix's ?revoke=true) beside the PUT/GET ones already documented.
- LinkedAccounts.tsx: a delete answering 404 means the account is
already gone, so treat it as the delete's end state — re-fetch and
close the dialog instead of showing an error.
- matrix_account.rs: matrix_logout treats a 401 M_UNKNOWN_TOKEN as the
token already being revoked and proceeds with the delete; every
other logout failure still keeps the account. Adds unit tests and
updates docs/swarm/ui.md to match.
Each row of an agent's linked accounts, except its own `main` matrix
account, gets a delete action. swarm-controller serves DELETE beside each
PUT (matrix-accounts/{account}, forge-accounts/{label}, github-account),
answers 404 for an account the store does not hold, refuses `main`, and
removes every version through `delete_all_versions`.
The matrix confirmation has a revoke checkbox, off by default: the
controller logs the stored token out at its homeserver first, and keeps
the account when that fails or no homeserver is stored.
The controller's policy gains `delete` on each agent's
`metadata/.../matrix/+`, `forge/+` and `github-token`, pinned in
bao-grants.nix.
Refs #4855
The detail panel makes one request for the agent it shows,
GET /api/hives/{hive}/agents/{agent}/linked-accounts, which returns every
matrix, forge and github account of that agent as names and hosts. The
all-agents route and the table's matrix-column rows are removed, so the
table makes no linked-accounts request. The panel stays keyed by
hive/agent. The bao grant is unchanged.
Refs #4855
GET /api/agents/linked-accounts returns one entry per agent that
/api/agents/status has a row for, as {hive, agent, accounts}, from one
store login. The agents page fetches it once (and again when a link dialog
closes) and hands each table row and the detail panel its agent's slice,
so the page makes no per-agent request. The per-agent route had no caller
left and is removed. The bao grant is unchanged: the same list on each
agent's matrix and forge metadata directories.
Refs #4855
Selecting another agent kept the previous agent's rows on screen until
the new fetch landed. Keying both mounts by hive/agent remounts the
component on an agent change; a version bump still refetches in place.
Refs #4855
GET /api/hives/{hive}/agents/{agent}/linked-accounts returns one row per
account linked to the agent, as kind, name and host: each matrix account
under swarm/agents/<agent>/matrix (with its homeserver, and the agent's own
`main` marked reserved), each forge label under swarm/agents/<agent>/forge
(with its url), and github when swarm/agents/<agent>/github-token exists
(host github.com, which is not stored). No credential field is in the
response type.
Listing those two directories needs a new controller grant: `list` on
secret/metadata/swarm/agents/+/matrix and .../+/forge only, pinned in
bao-grants.nix as the only metadata stanzas under agents/ beside the queue
revocation. Checked against a dev OpenBao 2.6.3: the grant lists those two
directories and is refused on agents/, agents/<agent>/, and a leaf.
The swarm UI agent detail panel shows all rows under "accounts"; the table
view's matrix column shows the matrix rows. The link badges stay.
Refs #4855
- docs/web-ui/README.md: drop the removed Credentials tile from the
H0M3 hub list.
- api-error.ts, hive-warn.js: rewrite comments pointing at
dashboard/src/credentials.js and credentials.html, now deleted, to
state what the code does instead.
- swarm-secret-client/src/github.rs: correct the Credential.value doc
to the actual read command (bao kv get -format=json | jq
.data.data.value), keeping the load-bearing-field-name point.
- github-token.nix, agent-github-bao.nix, LinkGithubAccountForm.tsx:
restate added comments as current behaviour instead of changelog
wording ("has always had", "holds the token now").
Refs #4347
An operator links an agent's GitHub personal access token in the swarm UI
(LinkGithubAccountForm, "link github account" on /agents). swarm-controller's
PUT /api/hives/{hive}/agents/{agent}/github-account stores it at
swarm/agents/<agent>/github-token (swarm_secret_client::github), a flat leaf
under the agent's prefix that the agent's existing read grant already covers:
no policy change, and no list grant, since there is one token per agent.
In the agent, hive-agent-github-token (oneshot + 2-minute timer, as the agent
user, under its own store certificate, ordered before hive-github-notify)
reads that path and writes <state>/github-token, 0600 and agent-owned, the
file the gh wrapper, git credential helper and hive-github-notify already
read. It replaces the file by rename only when the bytes changed and never
deletes it: a hive-written github-token stays until a token is linked in the
swarm UI. It is installed only with a store address and
services.hyperhive.agent.github.enable.
Removed: the dashboard's CR3D3NTIALS page (credentials.html/js/css, its
build entries and H0M3 tile; GITHUB was its only tab), hive-c0re's
dashboard/matrix_accounts.rs with GET/POST /api/github-account,
priv_client::write_agent_github_token, the host socket's
SetAgentGithubToken and `hivectl github set-token`, and hive-priv's
WriteAgentGithubToken with write_agent_state_file, its only caller gone.
Docs: integrations/github.md and swarm/ui.md describe the swarm path,
swarm/credentials.md gains the store-path row, and the hive UI docs,
hivectl docs and security.md's hive-priv table drop the removed pieces.
Closes#4347
An operator now links an agent's external forge account (label, base URL,
token) in the swarm UI. swarm-controller stores it at
swarm/agents/<agent>/forge/<label>. There is no index: the store's
listing of the agent's forge/ directory is the set of accounts.
In the agent, hive-agent-forge-accounts (oneshot + 2-minute timer, as
the agent user, under its own store certificate) lists
swarm/agents/<agent>/forge/ with the `list` #4866 grants an agent on its
own metadata subtree, reads each account, and writes
<state>/forge-<label>-token and forge-<label>.json in the names and shape
hive-forge -f already reads. An empty listing (a 404, which `bao kv list
-format=json` answers with `{}` and an empty stderr) is zero accounts; a
denial or an unreachable store fails the unit. It never deletes: files
for labels not listed, including ones the hive wrote, stay as they are.
Removed: the dashboard FORGES tab (credentials.js/html section and its
CSS), hive-c0re's extra_forges.rs and its routes, priv_client's
extra-forge calls, and hive-priv's WriteAgentExtraForgeAccount /
DeleteAgentExtraForgeAccount with their helpers. The GITHUB tab and
WriteAgentGithubToken stay.
Also: persistence.md's matrix avatar note names the exit-75 restart on a
changed account listing, not the dashboard, as what brings a linked
account up.
Refs #4348
hive-matrix-daemon now learns which external matrix accounts it has from
the swarm secret store, under the agent's own certificate, and the hive
push chain for matrix is gone.
The daemon lists swarm/agents/<agent>/matrix/ (the `list` its policy
grants on its own metadata subtree), reads each account's homeserver
from its credential, and brings the accounts up with their tokens from
the store. Every two minutes it lists again and exits with 75 when the
set of linked accounts changed; the unit restarts on 75 without counting
a failure. A listed name whose credential reads as absent is skipped and
logged once. At start it removes the matrix-token-<a> /
matrix-account-<a>.json pairs a hive delivered (a sidecar marks a pair
as delivered; a declared tokenFile keeps its token).
Removed: CredentialNotice and the $SWARM.credential.* subject and NATS
grant, the controller's publish and its queue precondition on the PUT
route, hive-c0re's credential subscription arm and workers/credential.rs,
priv_client::write_agent_matrix_token, hive-priv's WriteAgentMatrixToken
and its helpers, and the daemon's state-dir account discovery.
Kept: WriteAgentGithubToken and the external-forge path
(WriteAgentExtraForgeAccount, extra_forges.rs) are untouched, and a
declared matrixAccounts tokenFile is still read when the store has no
token for that account.
Refs #4348
mara, PR review: "the dynamic tab should be in the top bar, not a new
one below". Moves the .shell-tabs group from its own sticky row under
the header into .shell-nav itself, right after the nav indicator.
Also adds a third re-measure effect for the sliding nav indicator,
keyed on tabs.length: with tabs inline in the same flex row the
indicator measures, closing a background tab (no navigation) can
shrink the row without the hop effect's own re-measure ever firing.
Same reflow-not-navigation reasoning as the existing resize-listener
effect.
argus's review on PR #4784 caught a wrong technical claim: the
comment said the .ui-agent-term-preview-full override rules relied
on source order because they had equal specificity to the
un-modified rules above. They don't — each override selector adds
one more class (the .ui-agent-term-preview-full prefix) than what it
overrides, so they're strictly more specific and win regardless of
file order. Corrected the comment to say so.
Adds a full, non-capped agent terminal reachable from a new expand
trigger on the embedded AgentTermPreview (the detail-panel preview on
AgentsPage stays as-is, just gains the trigger). Opens
/agents/:name/terminal in a new dynamic tab in Shell's header, next to
the static nav row — tabs persist across a reload via useDynamicTabs,
a small localStorage-backed hook built on @hive/shared's existing
settings-storage primitive.
AgentTermPreview gains two new props to support both mounts from one
component: fullHeight (drops the 12em preview cap, fills its page)
and showHeaderBadges (default true — lets a future caller that
already shows turn_state/model/ctx/cost elsewhere suppress this
cluster; AgentsPage doesn't use it, see below).
Deviation from the originally posted plan (issue comment 80596): that
plan proposed AgentsPage's embedded preview pass showHeaderBadges as
false, reasoning the detail panel already duplicates that info.
Checked the actual code before implementing — it doesn't; AgentRow/
AgentTypes.ts carry none of turn_state/model/ctx/cost, and
AgentTermPreview's own floating badges are the only place swarm-ui
shows them. Left the badges visible there instead of shipping a
regression the plan's own stated justification didn't hold up to.
Also fixed a same-tab pub/sub race found by actually rendering a cold
load of /agents/:name/terminal (headless chromium, not just reasoning
about the code): useLocalSetting subscribes inside a useEffect, and
mount effects fire children-before-parents, so a descendant's
mount-time write (AgentTerminalPage registering its own tab) can beat
an ancestor's (Shell's) subscription into existence, leaving Shell's
tab row silently empty on a direct/reload load. Fixed by having
useDynamicTabs re-sync from storage on every location change, not
just on notify() — the fix lives in the new hook itself, not in the
shared settings-storage primitive theme/motion overrides also use.
Each agent card now leads with the agent's icon, loaded as an `<img>`
from `GET /api/agents/<name>/icon`: the same 5em square, background and
fallback as the hive dashboard's container row. An agent with no icon
(the route's 404), or any other failed load, shows the dimmed hyperhive
mark (`/favicon.svg`) instead of a broken image.
Only ever an `<img>`, never inline markup: the body is an agent-authored
SVG, and an image load does not run its script.
Header-bar parity for the swarm-level agent terminal, per mara's ruling
"dont make it a header though - make it floating elements on top of the
terminal". Adds useSwarmAgentStateStream (SSE consumer for
swarm-controller's GET /api/agents/{name}/state/stream) and renders a
small read-only Badge cluster absolutely positioned in
AgentTermPreview's terminal box corner, reusing the same anchor pattern
@hive/shared/terminal.css's .tail-pill already uses.
Shows turn_state always, agent_state only when it isn't the boring "up"
case, model (with resolved-model tooltip), and ctx/cost usage. No badge
renders until the first header event lands - the swarm queue's
agent-state subject is transition-only with no seed/replay, and no
swarm-level endpoint today can seed turn_state/model/ctx/cost
synchronously.
Vocabulary (state tones, token-total/format helpers) ported from
@hive/agent's Root.tsx/lib/format.ts rather than imported, matching the
existing "small helper, not worth a cross-package hop" call
useSwarmTermStream's own appendRow already made.
Verified by rendering the real component tree against a stubbed
EventSource and screenshotting the result - the badge cluster initially
overlapped the terminal's first row of text, fixed by giving this
preview's own .live.terminal extra top padding.
mara's steer on the issue, right as the previous commit's plan was
posted: the swarm-level term isn't a separate page, it's part of the
same pwa -- selecting an agent should show a small preview (no header,
no input) below the existing detail fields. That's the whole MVP for
this issue; the full terminal (+ sending input back to the agent) is
explicit follow-up scope.
AgentTermPage -> AgentTermPreview: drops the Panel wrapper, the
"terminal" nav badge + wouter route, and the back-link -- just the
inline preview, height-capped at 12em (@hive/shared/terminal.css's own
default is a much taller 72vh/60em, sized for a dedicated page rather
than a peek).
Moves the TermMsg rendering pipeline (Row.tsx, termMsg.ts, linkify.tsx,
markdown.ts) from @hive/agent into @hive/shared, so swarm-ui becomes a
second consumer of it instead of forking a copy -- CSS was already
shared (@hive/shared/terminal.css). marked+dompurify move from
@hive/agent's deps to @hive/shared's; swarm-ui picks them up
transitively, no new direct dep there.
New swarm-ui route /agents/:name/term (AgentTermPage), linked from
AgentsPage's detail panel via a "terminal" badge next to "link matrix
account". Consumes GET /api/agents/{name}/term/stream: unlike
@hive/agent's own useLiveStream (TermEnvelope-wrapped, history/backfill
dance), the swarm relay forwards one bare TermMsg per SSE event with no
envelope and no history endpoint -- useSwarmTermStream is a much
smaller hook for that shape (EventSource -> parse -> coalesce, nothing
to buffer/dedupe/backfill against).
Verified against a live SSE mock (screenshots in /agents/iris/state/screenshots/
3801-agents-detail-panel-terminal-badge.png and
3801-agent-term-page-live-rows.png -- real rows rendering through the
shared Row component, not just a build/typecheck pass).
Ports the subtle text-shadow glow @hive/agent's and the dashboard's own
headline rule already use (both independently converged on
`text-shadow: 0 0 Npx color-mix(color, transparent)`), not their whole
uppercase+letter-spacing headline typography -- that's those UIs' own
look, swarm-ui never adopted it and this issue only asked for the glow.
- Panel titles: `currentColor`-based, since a panel title has no one
fixed accent (some are dynamic, e.g. an agent's own name) -- a fixed
--purple glow on plain --fg text would read as a mismatched halo.
- Active nav tab: each item's own `accent` (NAV_ITEMS), the same value
its underline already uses, so the glow and the underline read as
one accent rather than two that happen to agree.
AgentsPage's table view (and its list+detail split) hits the same
60em readable-line-length cap the issue report's wide table already
opts out of via shell-body-wide. Adds /agents to WIDE_BODY_ROUTES --
the allowlist's own comment already anticipated this, a future wide
page reaches for the same class rather than inventing its own cap.
Dialog and Panel both drew their own bordered/backgrounded card in the
same --bg-elev, so a Panel-based dialog (create-agent, link-matrix-account)
rendered as two concentric cards with a floating close button on the
outer one and no purpose to it.
Give Dialog a "plain" mode that drops its own card chrome (border,
background, padding) and floating close button, and give Panel an
optional onClose that renders a close button at the end of its own
header row instead. AgentsPage's two Panel-backed dialogs now use
plain + Panel's onClose, so the Panel is the dialog's only visible
card. ConfirmDialog (no Panel of its own) is unaffected — plain
defaults to false, unchanged card + floating close button.
Added a ComponentsPage sample demonstrating the plain + onClose
pairing. Verified both dialog modes via a real headless-chromium
screenshot (plain dialog: single card, close button in the header
bar; default dialog: unchanged floating close button).
mara flagged (PR #4297) that the comment-block lint doesn't catch this
file at all -- check-comment-blocks.sh (and check-issue-refs.sh) scope
to a fixed git ls-files extension list that omits .mjs. That's a real
gap (dashboard/build.mjs already has a 49-line block, pre-existing),
tracked separately -- but the fix here is to just respect the 30-line
convention regardless of whether the lint enforces it on this
extension. Cut the redundant per-file output-layout table (the code
below already names every output path) and kept only the genuinely
non-obvious decisions -- no content lost, verified against a plain awk
run of the lint's own line-counting logic.
Every redeploy currently changes the nix store path serving swarm-ui's
JS/CSS but never the URL the browser requested (main.js, main.css,
theme.css, swarm-ui.css are all fixed filenames) -- so a browser can get
stuck serving yesterday's bundle after a deploy until someone clears the
cache by hand.
build.mjs now hashes main.tsx's JS bundle + its companion CSS output,
plus theme.css and swarm-ui.css, via esbuild's own metafile (not a
guessed hash algorithm), and rewrites the real URLs into index.html and
sw.js after the build.
colors.css deliberately stays unhashed: nix/host-modules/swarm-ui.nix's
stylix theming swaps that exact fixed path via an `= /static/colors.css`
nginx location override with no npm/esbuild rebuild involved. Hashing it
would silently break that swap on a themed host.
sw.js's CACHE_VERSION is now derived from the real hashed URLs instead of
a hand-bumped literal, so any shell-asset change gets a fresh cache name
and the SW's own activate-time sweep evicts the previous one in full --
fulfilling a promise its own prior comment already made.
Scope: swarm-ui only, per mara's call on hyperhive#4283 ("this is about
swarm ui - everything else will eventually migrate over"). dashboard and
agent are untouched.
The response.ok fix from the earlier review round was never reformatted
after editing. nix flake check has no --keep-going, so this one real
treefmt-check failure was aborting the whole check run -- and whatever
other derivation happened to be mid-build in the same parallel batch got
killed and logged with an empty error message, misread as an
infrastructure flake across many CI attempts. No logic change.
- manifest.webmanifest: mara — "name should be hyperhive", was
"hyperhive swarm".
- sw.js: argus — cache-write path didn't check response.ok before
caching; an error response would get served back as if it were the
real asset on the next offline/failed fetch.
Closes#4282. mara: "scope looks good" — approving the plan posted
there (manifest + icons + minimal shell-only service worker + iOS meta
tags) and both explicit questions (network-first-with-offline-fallback,
never cache /api/*).
docs/web-ui/design-guide.md's "Layout & viewport" section already
asserted swarm-ui is installable as a PWA — this is what actually backs
it.
- manifest.webmanifest: name/icons/start_url/standalone display, theme
#cba6f7 / background #1e1e2e matching the mocha --purple/--bg values.
- sw.js: plain JS, not TypeScript — the DOM lib swarm-ui's own tsconfig
uses and the WebWorker lib a service worker's globals need are
mutually exclusive in one tsc program, not worth a second tsconfig for
a self-contained ~100-line file. Scoped to the app shell only, never
touches /api/* at all, network-first with offline-fallback-to-cache
(not cache-first) since main.js/main.css are unhashed filenames and a
cache-first SW would risk wedging an operator on stale JS after a
deploy.
- index.html: manifest link, theme-color meta, iOS
apple-mobile-web-app-* tags (Safari ignores the manifest spec).
- main.tsx: feature-detected SW registration.
- branding/hyperhive-maskable.svg: hyperhive.svg's own artwork already
fills nearly its whole canvas, so a maskable icon needs a padded
variant or an OS mask crops the outer ring/corner brackets — embeds
the original via a scaled <image> ref rather than duplicating markup.
- nix/packages/swarm-ui.nix: rasterizes icon-192/512/512-maskable.png
from the branding SVGs at build time via librsvg, rather than
checking in static PNGs.
Verified for real: typecheck+build green, and a real headless-chromium
tab driven over CDP confirms the service worker registers and becomes
the active controller, and a simulated-offline reload still serves the
full cached shell rather than a browser error page. nix build .#swarm-ui
also verified green, including the rasterized icon output.
mara, PR review: "make agentspage a subdir now that its split into
sub components". AgentsPage.tsx/.css, AgentCard.tsx/.css,
AgentTypes.ts, and WantedMenu.tsx move as a family into their own
pages/agents/ directory; CreateAgentForm and LinkMatrixAccountForm
stay in pages/ since they aren't part of this split (CreateAgentForm
is still rendered inside AgentsPage's own dialog but is a standalone,
independently-named form, not one of the pieces carved out of the
page itself).
Pure move: relative imports within the new pages/agents/ family are
unchanged (they were always siblings), only the ones reaching back
out to ui/ and the two forms above gained one more '../', plus
App.tsx's route import.
Three more asks from the same review thread:
- "agentspage is now giant and deserves a split" - AgentsPage.tsx was
1047 lines. Split into AgentTypes.ts (AgentRow and friends),
WantedMenu.tsx, AgentCard.tsx (+ its own CSS), leaving AgentsPage.tsx
as state/actions/columns/the render tree - 649 lines, and every piece
it composes is now independently readable.
- "what about the component that represents filtered data ... that the
card view and table can both use?" - extracted FilterableView
(ui/filterable-view/): takes columns/rows/rowKey/storageKey/view/
renderCard, builds its filter bar from *every* filterable column (not
a hand-picked subset - the old AgentFilterBar only showed 4 of the
agent columns' 6 filterable fields, an accidental gap the table's own
popovers didn't have), and renders either the card list or Table.
AgentsPage now just tells it which view to show; the view toggle
itself stays page-side since it's Panel-header chrome, not filtering.
Disclosed side effect: card view's filter bar now also covers
message/config-PR (text filters), matching table view exactly instead
of a narrower subset.
- CSS audit: AgentsPage.css now holds only what's genuinely page-specific
(the view toggle, the detail-panel field grid) - everything else moved
to its owning component's own colocated CSS.
FilterableView gets a /components demo (view toggle + filter bar + both
render modes, same day per the design guide). Verified: AgentsPage
still renders the same (real screenshot), and the demo's own table
toggle produces a real Table with the same rows.
argus, PR review: detailTarget stored the whole AgentRow object at
selection time, so it never picked up a later refresh() or a
declareState patch - the card list updated live, the panel next to it
kept showing whatever was true the moment it was opened. Concrete
repro: open an agent's detail panel, wait for the next refresh or
destroy it from inside the panel itself, watch the panel not update.
Fix: store only the selected agent's name (detailTargetName) and
re-derive the actual row from rows on every render
(rows?.find(r => r.name === detailTargetName)). The panel can't drift
from what the list is showing since it's reading the same array.
Verified with a real refresh cycle against a mock server that returns
different data on the second call: before, both card and panel show
"idle"; after a live refresh, both show the new value in sync.
