A new `swarm-builder` service, in its own container on the host that sets
`deploy.builder.enable` (defaulted on by `allSwarmServices` where the host
has the builder's store leaf, like the review classifier).
When an `agent-configs/<agent>` PR is opened, reopened or synchronized,
swarm-controller queues a `build_agent` jobq node. The node sets a
`swarm-builder` commit status to pending on the PR head, sends
`{agent, rev, pr}` to the builder over NATS request-reply
(`$SWARM.builder.build`), and writes success (store path), failure (last
line nix printed) or error (no builder / no reply) on the head. The node
fails when the build does, so the Jobs page matches the PR.
The builder builds nix/builder/check-flake's `check` with every input
overridden: hyperhive and nixpkgs to the host's own
`c0re.hyperhiveFlake`/`c0re.nixpkgsFlake` (what a hive there renders into
its meta flake), and the agent to the PR head on the forge. Evaluation is
pure and runs in the container; builds go to the host nix daemon as an
untrusted client. The private config repo is fetched with the token of a
new read-only `builder` forge account (in the `hives` team), which the
controller mints into `swarm/services/builder/forge/token`; git gets it
from a credential helper reading the child's environment, so it is in no
argument or URL.
Credentials: a `builder` store leaf minted by glue-bao-tls on the store
host, a `swarm-builder` reader policy (forge token + its own OIDC client
secret), an authelia machine client, and a `--builder-client` grant in the
queue's auth-callout responder. The controller gains a grant on the token
path and the secret publisher a deny on it.
This is the agent-config build check atlas scoped on #5081 (slice 3):
nothing is pushed to a cache, nothing is written to KV, and a red status
gates no deploy. The check composes agent-base + the config without the
values a hive adds (CA, docs, claude-code path, ports, input follows), so
it can disagree with a hive's own build.
Add docs/swarm/review-classifier.md, the high-level architecture of the
planned PR review classifier pipeline, written as the full target shape
under a "planned, not implemented yet" banner. Link it from docs/README.md
and the docs/swarm/README.md piece list.
The page reflects the epic description and the operator's answers in the
epic thread; points still open are listed in its "Still open" section.
The banner names the epic by title rather than by number because the
tracker-tag lint rejects issue references in tracked files.
swarm-controller keeps each agent's inference settings in a new
`agent-inference` KV bucket, one key per agent, and is its only writer
(`GET`/`PUT`/`DELETE /api/agents/{name}/inference`; `swarmctl agent
inference get|set|clear`). Every field is optional. Each agent reads its
own key every 10 s through a direct-get grant on that key alone, and a
set `model` or `effort` replaces the Bus value, so the next turn runs on
it and the badge follows.
The swarm value is applied without persisting it: `Bus::set_model` and
`set_effort` are split into a persisting setter and a non-persisting
`apply_*`, and the poller uses the latter. Otherwise the swarm value
would overwrite `hyperhive-model`/`hyperhive-effort`, and clearing it, or
a restart, would no longer find the agent's own value. An unset field
changes nothing; a field cleared after being applied goes back to the
local value (`harness_state::local_model`/`local_effort`, the same
resolution `Bus::new` uses), read again at the clear.
`subagent_models` is validated, stored and shown by `get`; no agent
applies it yet.
Refs #5034
The CI runner was a hive deploy module: hive-ci.nix ran it beside the
forge, and hive-c0re registered it through the forge's admin API with
the `core` token, writing the registration token through hive-priv. It
could only run on the forge's host, and it kept an admin-scoped forge
token on a hive.
Now the runner is a swarm service:
- swarm-ci-runner.nix runs it in a `swarm-ci-runner` container on the
host that sets deploy.ci-runner.enable, reaching the forge by name.
It needs neither the forge nor a hive there.
- swarm-controller reads the forge's runner registration token with its
own admin token and writes it to swarm/services/ci-runner/
registration-token whenever it differs from what it last wrote. Its
grant on that path is create/update only.
- The runner's own leaf (CN swarm-ci-runner, role and policy of the same
name) reads that one path. swarm-bao-ci-runner.service fetches it into
the runner's EnvironmentFile and restarts the runner on a change.
- A pre-start step removes .runner when the daemon's journal since its
last start holds Forgejo's "unregistered runner" refusal (the daemon
exits 1 when its startup Declare gets it), or when .runner names
another forge address; upstream's pre-start step then registers again.
Removed from the hive side: hive-ci.nix and deploy.forgejo.ci.*,
forge/ci_runner.rs, PrivRequest::RegisterCiRunner and its handler and
ReadWritePaths entry, InfraContainer::Ci with hivectl's --ci and the
dashboard's hive-ci journal entry, HYPERHIVE_FORGE_CI_ENABLED, and
read:admin,write:admin from CORE_TOKEN_SCOPES: nothing else in hive-c0re
calls /api/v1/admin/*. The old ci options, under both swarm.forge.ci and
deploy.forgejo.ci, now fail evaluation naming deploy.ci-runner.
The forge's half of the old ci.enable, the actions/checkout mirror and
DEFAULT_ACTIONS_URL = self, is deploy.forgejo.actionsFromSelf.
Closes#4954
config-changes.md (was approvals.md) describes config PRs, the meta flake
and helper events; the approval queue, its kinds, the withdraw path, the
job-queue dispatch and ApprovalResolved go. Every other page drops the
approvals table, endpoints, event, tool group, hivectl verbs, dashboard
section and the loose-end/cancel kind, and links to the renamed page.
Drops docs/tools/scheduling.md and every description of hive-level
schedule creation, the worker, the removed MCP tools, the scheduling tool
group and the removed endpoints/events, pointing at docs/swarm/schedules.md.
swarm-controller keeps every scheduled prompt (id, name, agents, body,
anchor, interval, paused) as one document under one key of a new
`schedule-definitions` KV bucket, so each change is a single
compare-and-set and a prompt exists for all of its agents or none. It
writes each agent's key of a new `agent-schedules` bucket as the prompts
naming that agent, without names or other agents, after every change and
on a one-minute reconcile pass. It is the only writer of both, behind
GET/POST /api/schedules and PATCH/DELETE /api/schedules/{id}.
`swarmctl schedule add|list|edit|pause|resume|rm` drives those routes over
the controller socket.
An agent's own queue credential is granted one more subject, the direct
get of its own key in `agent-schedules`. A KV watch is not granted: its
consumer create subject carries no key, so it would cover every agent's.
hive-agent's new schedule ticker reads that key every 10 seconds and fires
due slots as the same `schedule:<id>` todo hive-c0re's worker pushes. Fire
times are anchor + k*interval; the last fired slot per schedule id is kept
in the state db, so a body edit or list refresh never moves or refires a
slot, downtime fires the latest missed slot once, a restart refires
nothing, a one-shot fires once, and paused slots are skipped.
Refs #4914
Per mara's #4879 review (89815): the system has no agent hierarchy,
just a flat set scoped by the capability store, so the filename no
longer matched. The file already read "Agent roster & privileges"
after the earlier facts pass; rename it to match, and update the
four inbound references (docs/README.md, coordinator.md, config.md,
agent-service.nix).
Fix the 4 pre-existing vale error-level hits on main (docs/README.md:83,
docs/getting-started/setup.md:11,127, docs/swarm/bao.md:4) that fail CI's
prose-lint-errors job for every docs PR regardless of its own diff.
The topology doc keeps its filename and its second half (manager
special-casing, harness unit shape) — both are cross-referenced from
other pages and neither is about the parent field. Its first half is
rewritten: what topology.json is now, and a table of what the removal
took with it, so a reader who finds `<parent>` or `set-parent` in an old
issue thread learns it went away rather than moved.
The dashboard's tree-rendering section is marked dormant rather than
deleted: the walk is still in swarm.js and retiring it is the frontend
owner's call.
Fixes the Microsoft.Contractions/write-good.ThereIs/alex.Condescending/
Microsoft.Auto errors vale flagged in CI (prose-lint-errors, required):
contractions throughout, no sentences opening with 'There is', dropped
the 'auto-mints' hyphenation and the 'easily' condescension flag, and
renamed the credentials table's 'what it is' header to 'kind' (also
flagged once styles were resynced). No technical claim or file:line
citation changed. Same fix applied to the one new line in
docs/README.md's pointer to this page.
The swarm's credential docs say where every file lives. They do not say
whether it should be a file at all, so a discussion about direction has
had nothing to point at and each one re-derived the same table.
This page carries that table with the three columns the target contract
is written in — minter, reader, renewal — plus the column the target is
really about: whether the value is persisted outside the store. Stating
it flatly is the point. All four stored families are plaintext files on
disk, the appservice token twice; every renewal cell reads NONE; no agent
container holds a store identity at all, so the hive reads on its behalf
and writes a file in; and the appservice token has a second, uncoordinated
local mint that can diverge from the published one.
The target section is marked as a target throughout, because its first
line is the one most easily misread as fact: every host needing a store
mTLS certificate is where this is going, while today only the store's own
host auto-mints and swarm-bao.nix calls it the credential an operator
places by hand everywhere else.
The progressive-enhancement rule is stated as a table of questions a
reviewer applies to a pull request rather than as prose, since a rule
nobody can check is a preference. New functionality matches the target
immediately; existing functionality moves stepwise, and the questions
distinguish a step from churn.
Indexed from the docs root and the swarm README. It supersedes
secrets.md when the migration completes — at which point that file is
deleted and this one moves into its place.
swarm-controller's `InitAgentConfigRepo` node already covers config-repo
creation, so this deletes a duplicate rather than a capability; old
`init_config` rows are skipped by `collect_lenient` with no migration, by
operator decision.
Refs #4398
Fixed 72 hyphenated 'auto-X' compounds across the docs by deleting
the hyphen to match Microsoft style guide (autogenerated, autoregistration,
autodiscovery, etc). Preserved original capitalization throughout.
Covered 26 documentation files in docs/, including README.md, agent-lifecycle,
integrations, scheduler, networking, process, tools, trust-boundary,
turn-loop, and web-ui sections.
Review pass on top of the previous commit (85a94671) turned up three
real issues its own vale-only verification didn't catch:
- 'has not' -> 'hasn't' fired inside 'has nothing', corrupting
docs/swarm/README.md into 'hasn'thing to say upward'. Restored
'has nothing'.
- ~44 hits where the original text was genuinely sentence-initial
('It is', 'That is', 'They are', 'What is', 'Do not', ...) got
contracted in lowercase, silently de-capitalizing the start of a
sentence. Checked each one's real surrounding text and capitalized
the ones that needed it.
- Confirmed (by reading the actual text, not just trusting vale) that
the 3 remaining 'is not'/'has not'/'it is' hits after those fixes are
vale's own false positives: it matches these as a bare substring with
no trailing word boundary, so it also fires on 'is nothing', 'has
nothing', and the already-correct 'it isn't'. Left those alone --
there's no real issue in the text.
Re-verified with a fresh vale run: the remaining 17 hits are exactly
the union of the 14 legitimate ALL-CAPS/bold-emphasis skips from the
first commit and these 3 false positives. Full accounting in
/agents/iris/state/haiku-batches/contractions-report.md.
Apply contraction fixes across ~40 doc files (setup, integrations, lifecycle,
networking, scheduler, swarm, tools, trust-boundary, UI, etc.). Skipped 14 hits:
- 10 where words appear in ALL CAPS for deliberate emphasis (is NOT, do NOT, etc.)
- 4 where text could not be safely located due to markdown formatting or column position
Applied via systematic scan with checks for fenced code blocks, inline code spans,
and intentional caps. Preserves sentence-initial capitalization throughout.
hyperhive#3902, mara: option (a) - a content pass splitting mixed docs
into operator-facing content plus implementation detail. One file
first, to agree on the split pattern before doing the other ~19.
Went through 3 shapes on review before landing here: a sibling
-internals.md file (mara: clutters the navigation), then two
tree-precedent alternatives damocles raised (subdir+README like
web-ui/; or only split docs with a pre-existing boundary marker,
which would've covered 3-4 of the ~20 flagged docs and left the rest
untouched), then mara's own proposal - a collapsed <details> section
in the same file. Verified empirically (cmark-gfm --unsafe, the
website's own render pipeline) that markdown headings nested inside a
<details> block still parse as real headings with heading-id anchors
once separated from <summary> by a blank line, so anchor links into
the collapsed section keep working.
github.md keeps enabling/provisioning/security up top; its
'Implementation' section is now a <details> block holding what was
briefly a separate github-internals.md (deleted again) - how the
agent's gh/git-push actually authenticate, and the notification
poller's internals. Reverted the two cross-references + the
docs/README.md entry back to pointing at github.md now that the
content lives there again.
Added a short CLAUDE.md note recording the pattern per mara's ask,
including the one real caveat damocles flagged: <details> only
collapses in a rendered browser, a raw-text read (cat, the Read tool)
still sees everything, same as today.
Per mara's go-ahead on hyperhive#3902 ("getting started is good, but
terminal rendering does not go in there i think"):
Moved 21 top-level docs/*.md files into 7 new topic subdirectories
(existing web-ui/, turn-loop/, swarm/, tools/, crates/ untouched):
getting-started/ setup.md
agent-lifecycle/ agent-hierarchy.md, approvals.md, persistence.md
trust-boundary/ boundary.md, security.md
integrations/ forge.md, matrix.md, github.md, knowledge.md
networking/ gateway.md, network.md, snapshot-store.md
scheduler/ jobq.md, coordinator.md, ci.md, observability.md
process/ conventions.md, gotchas.md, pr-review-gate.md
web-ui/ terminal-rendering.md (moved into the EXISTING dir,
per mara's correction to the original getting-started
guess -- it's UI implementation detail, not onboarding)
The physical layout now matches docs/README.md's own topical headers,
which already amounted to this taxonomy -- see the scoping comment on
the issue for the two findings that motivated this (a genuine
duplication between CLAUDE.md's old "Reading paths" list and
docs/README.md's grouped one, since drifted out of sync with each
other; and the flat layout not matching the grouping we already had).
Fixed every cross-reference this moved across the whole repo (~120
files: docs/ internal links at every depth, Rust doc comments, nix
module option docs, crate READMEs) -- verified two ways: a grep sweep
confirming zero remaining references to any old path, and a script
that resolves every markdown link in docs/**/*.md + CLAUDE.md +
README.md against the filesystem and reports anything that doesn't
exist (zero broken links).
Collapsed CLAUDE.md's "Reading paths" section (the duplicate) down to
a pointer at docs/README.md, now the single index. Rewrote
docs/README.md itself to use the new subdirectory paths and added the
one doc it was missing that CLAUDE.md's old copy had (pr-review-gate.md).
Classified all 22 docs/*.md files first via a haiku subagent (mara's
suggestion) on two axes -- proposed grouping and operator-vs-
implementation focus -- before finalizing the taxonomy; spot-checked
the report and found internal inconsistencies (its classification
table disagreed with its own summary section for a few files), so this
taxonomy is my original proposal + the one correction mara gave
directly, not a blind application of the subagent's table. The
operator-focus data it gathered is still useful for a follow-up
content pass (docs skewing 'mixed' rather than pure operator-facing),
not addressed in this PR -- structure only.
nix fmt clean, both pre-push lints clean.
mara: "only the abstract jobq part was asked for in the first place" —
the previous revision still carried the full hive-c0re step-kind table
under a "core-specific nodes" heading; that catalogue belongs in
coordinator.md (where it already lived) alongside the rest of
hive-c0re's job-queue internals, not duplicated here.
jobq.md is now just the domain-agnostic model (graph of steps + shared
resource slots) plus the generic row/step/glyph framing for watching it
in the dashboard — no hive-c0re-specific step names anywhere.
Coordinator.md's job-queue section, docs/README.md, and the root
CLAUDE.md reading-paths index are updated to match.
hyperhive/hyperhive#2999 asked for an operator-facing page explaining
the job queue's node kinds in the abstract, without the coordinator's
implementation detail (leases, resource windows, module layout).
Add docs/jobq.md: what a queue row/step means, a plain-language table
of every node label an operator actually sees in the dashboard
(BU1LDS/R3BU1LD QU3U3 and swarm-ui's /jobs page both render the same
graph), and a rough shape for a typical rebuild.
Wire it into docs/README.md and the root CLAUDE.md reading-paths index
alongside coordinator.md. Add a pointer from coordinator.md's job-queue
section to the new page so the two don't try to explain the same thing
twice at different levels.
The generated landing page's own H1 already said "Crate reference" —
the directory name should match. Rename docs/components/ -> docs/crates/
and update the generating derivation (nix/packages/reference-docs.nix),
its default.nix caller comment, and docs/README.md's link.
Fixes#3193
Each workspace crate's own README.md now also shows up as
docs/components/<crate>.md via nix/packages/reference-docs.nix, so it
renders on the docs site alongside the rest of docs/ without a second
hand-copied file to keep in sync. Relative links that were correct
from the crate's own position in the tree are rewritten for their new
virtual position (../docs/x.md -> ../x.md, ../<sibling-crate> ->
./<sibling-crate>.md). Landing page at docs/components/README.md, nav
bullet added to docs/README.md. No changes needed in the website
repo's docs.nix - it already walks every subdirectory generically.
`docs/swarm.md` becomes `docs/swarm/README.md` and the shared-services
material moves to `docs/swarm/services.md`, following the shape
`docs/turn-loop/` and `docs/web-ui/` already use. The README keeps a
pointer so the reading path is unbroken.
Every referrer moved with it — five docs pages, two option descriptions
in swarm.nix, and CLAUDE.md's reading path. A pointer to a file that
moved is worse than one to a file that was deleted: the content still
exists, so the reader concludes the note is wrong rather than the path.
Fixes hyperhive#3014.
docs/README.md is a genuine hand-written index for the docs/ tree -
task-oriented reading paths grouped by topic, covering every top-level
doc plus the three subdirectories that already have their own landing
page (web-ui/, tools/, turn-loop/). Adapted from CLAUDE.md's existing
"Reading paths" section (already curated and kept current) rather than
written from scratch, reorganized into headed groups since this is a
landing page, not a flat reference list. Explicitly covers
docs/tools/matrix.md and docs/github.md, per the note on hyperhive#3014
about the content that moved out of README.md's deleted section on
PR #3006 not going undiscoverable.
Top-level README.md's reading-path table (enumerating every doc)
replaced with a single prominent docs-site link, per mara's "replace
docs links table in readme md with a prominent link to the docs
(public host and git relative path)" - both forms present (rendered
site URL, git-relative docs/ path).
Verified every link in docs/README.md resolves to a real file (33
files checked). nix fmt clean, tracker-tag grep clean.
Companion to hyperhive/website#47/#48 - this is what that PRs "index.html
half" needs to exist before it can render, per atlas's note on #48.
