GET /api/agents/{name}/bash-tasks reads the agent's keys of the bash-task-status
bucket and reports each task's state, command and times, with a live task whose
daemon stopped reporting as stale. The swarm UI shows them in a background tasks
section of the agent card, running ones first. The task output stays unread.
Nothing deletes the key of a finished task, so the bucket would grow with every
task an agent runs. The shared bucket helper takes an optional max age, applied
when the bucket is created; a live task's republishing keeps its own key.
A status key per task in the bash-task-status bucket and the task's output rows
in a per-agent stream, kept apart as for subagents. The controller creates the
bucket and the streams, and the agent credential may publish its own keys and
subjects. Nothing publishes yet; the bash MCP half follows.
The builder, ci-metrics and review-classifier forge accounts get an avatar: an
icon each in a hexagonal frame, the hyperhive logo when one is missing. The
controller sets them with a short-lived write:user token it mints for the
account, since only an account can set its own avatar.
An agent with no icon of its own shows a default agent icon in the swarm UI, a
robot head in a round frame, instead of the dimmed hyperhive mark.
One key per subagent (<agent>.<subagent>) holding running/finished, since,
exit, model, effort and the publish time. The controller creates the bucket,
and the agent credential may publish its own keys. Nothing reads or writes
it yet; the daemon publisher and the controller list follow.
The hive directory and the WireGuard mesh become handbook pages of their own,
the controller's mint-failure handling moves to controller.md, and the
credential-by-credential mechanics and the secret store UI's proxy move to a new
implementation page.
Each service section lost its container, credential and metrics detail to a
Host setup section on its own implementation page, unchanged apart from links.
The four agentAuth option rows that sat above the page title moved into the
telemetry page's metrics option table.
A weekly swarm-archive-check unit runs restic check --read-data-subset on every repository on the archive host, serialised with the prune by a shared flock. It logs archive check failed per agent, pushes check gauges, and has an alert and dashboard columns.
The document's disabled list takes archetype:<archetype>/<flow> (off for every member) and the same with @<agent> (off for one member); a key naming nothing is ignored. GET /api/notify/flows lists the archetype flows with their members, the board draws a card for each with a switch, a checkbox per agent and duplicate, and the dry run counts their hits. The flows keep following the archetype.
ArchetypeMeta gets flows, checked when the controller stores the record: every todo must go to the new members recipient, which the router expands to the agents of the archetype. The operator's flows document refuses members. The router reads the archetype records and each roster agent's archetype key, and its NATS grant gains read on both buckets.
The classifier's Anthropic and ACP provider keys were written by hand in the
secret store's UI. The controller now stores them: GET
/api/review-classifier/keys says whether each is set and when it changed, and
PUT /api/review-classifier/keys/{runtime} writes one. No route answers a key.
The controller's store policy gains create and update on the two leaves with
no read of their values, and a metadata read for the set state. The swarm UI's
classifier settings card gets a key form beside the settings, saved apart
because it goes to the secret store.
A domain variable, the destination host from the proxy's access log, sits beside
the agent one and every query on the board filters on both, so one domain and
its agents show which agents queried it and when.
The queue subjects, scratch directory and stdio steps of a login, the harness's
newer-version rule, the minter's grants, the caveat and the metrics move to
4-implementation/swarm/claude-login.md. The handbook page keeps how to log an agent
in, the stored login and who may do it. Design pointers in the nix module, the
crate README and AGENTS.md name the new page.
The page-by-page and control-by-control description of the swarm UI moves to
4-implementation/web-ui/swarm-ui.md, with the quick-links wiring. The handbook
page keeps the route table, how to reach the UI, who may and what the browser
keeps. Links to the moved sections point at the new page.
A todo node's key and text were hand-expanded `{field}` strings. They are now Jinja templates rendered through the new hive-template crate (minijinja, an unknown variable is an error), and the controller's save check reads the variables the template uses. The flows document is version 2; a version 1 document reads as version 2 with each `{field}` converted to `{{ field }}`, and the next save writes it.
credentials.md drops the target-shape framing and keeps the rules, where a
secret lives, the swarm- and hive-level secrets and how to add one. The queue
responder's keys, per-principal identities, how a reader reaches the store, the
store UI's login and where the root lives move to
4-implementation/swarm/credentials.md. The per-secret reference moves to
4-implementation/swarm/credentials-reference.md. Links and comments follow.
The operator parts of the persistence and config-changes pages merge into
agents/persistence.md. The sqlite tables, state dirs, mounts and migrations move
to 4-implementation/agent-state/persistence.md, and the deploy path, meta flake,
deploy markers, startup migrations, root container, system messages and
autoupdate move to 4-implementation/agent-state/config-changes.md. Links, code
comments and nix comments that named the old sections point at the new pages.
Account minting, knowledge sync, the archetype refresh, the matrix appservice
flow and the SSO secret delivery move to pages under 4-implementation/swarm.
Each handbook page keeps what an operator does or sees, with a pointer.
failures_only (off by default, also for a stored document without it) makes the
review the classifier posts on a PR list just the rows that failed their rule,
or say that no check failed on an approval. A review with no clean verdict and
a try-out keep every row. The classifier settings form gets a checkbox for it.
