The OTEL resource attributes are baked into every agent's
managed-settings.json at evaluation time, but claude-settings.nix read
the names from the container's environment.variables - where they are
never set. meta.rs forwards them as runtime env only, so the reader hit
its "unknown" fallback and every agent shipped
service.name=hyperhive-agent,agent=<a>,hive=unknown,swarm=unknown
on every metric, while the same process's env held pr1ma/constellation.
Measured in this container's /etc/claude-code/managed-settings.json.
The map from forwarded env var to agent option already existed for the
service URLs, with a doc comment naming this exact hazard: "setting only
one leaves the other on its default". The names were simply never added
to it. They are now, and the constant is renamed FORWARDED_VAR_OPTIONS
since it no longer holds only URLs.
hyperhive.hiveName / hyperhive.swarmName follow the forge.url shape:
nullOr str defaulting to null, where null means the hive did not name
itself and "unknown" is an honest label rather than a guess baked at
eval time.
Also fixes, unasked: CLAUDE_REMOTE_CONTROL_SESSION_NAME_PREFIX reads the
same value, so remote-control sessions were named "unknown-<agent>".
Gated with nix, not just cargo - the blast radius here is module eval,
which fmt/clippy/test cannot see:
agent-base assertions -> [] (no failures)
extendModules with both options -> hive=pr1ma,swarm=constellation
Note the value is baked, so every agent needs a rebuild before the new
label appears on its metrics.
With the agent option nullable, a missing `HIVE_FORGE_URL` would no
longer fail anything — it would deploy a whole fleet of agents that
silently never log into the forge. The forge is not optional on a
running hive, so the hive asserts that itself rather than leaning on a
module that legitimately allows "no forge" when evaluated standalone.
`sync_agents` checks it before writing anything. That is the moment the
hive commits to a flake, and it keeps `render_flake` a pure string
operation: the renderer is exercised directly by a dozen tests, so
making *it* env-dependent would force each of them to either set a
process-wide var — the parallel-test race this module already avoids —
or fail for reasons unrelated to what they assert.
`require_service_urls` is pure over the already-collected pairs, so its
two tests need no process env at all.
Refs #2860
Step 1 of removing the localhost fallbacks: make the renderer emit the
value it already knows, so the option stops being a second, disagreeing
source of truth.
These options existed but nothing ever set them, so every agent fell
back to their localhost:<port> defaults while the real value reached
the container only as an env var. The two are consumed at different
times — the option is baked into scripts at build time (tea-login's
FORGE_URL), the env var is read at runtime — so which answer a given
code path gets depends on which one it happens to read.
Emitting them here follows the shape the otel block already uses: host
state becomes build-time agent module config. It is the precondition
for deleting the defaults, which is the actual fix: a loopback address
is only correct when the callee shares the caller's netns, and the
forge and homeserver are moving to swarm level, possibly onto other
hosts.
An absent var emits nothing rather than a guess. Once the defaults are
gone that surfaces as an eval failure, which is the point — better a
build that stops than an agent quietly talking to a port on the wrong
machine.
The emit is a pure helper rather than an inline loop so it can be
tested without process env. The first version of the test set env vars
and rendered the whole flake; it failed because the parallel runner
raced it against the existing env-mutating test, not because of any
defect. Testing the pure function has no such hazard, and the
render-level variant is kept #[ignore]d with that reason recorded.
mara on PR #2769: "make the default null instead of special casing """.
`claude_code_path` was a `String` whose empty value meant "no host-level
pin". That is a sentinel doing an `Option`'s job — the same shape argus
and mara already rejected on #2755's weights, and the same
empty-field cruft mara called out on #2756.
So it is `Option<String>` end to end:
- host module: `claudeCodePath` evaluates to `null` when
`claudeCodePackage` is unset, so `serve.json` carries JSON `null`
rather than `""`.
- `Coordinator` + `HiveEnv`: `Option<String>`, defaulting to `None`.
- `render_flake`/`render_flake_with_lookup`: `Option<&str>`, and the
emission is an `if let Some(path)` instead of an `is_empty()` guard.
- agent module: `hyperhive.claudeCodePath` is `nullOr str`, default
`null`.
Behaviour is unchanged in both directions; only the way "unset" is
spelled moves. The `builtins.hasContext` assertion still guards the
pinned case (short-circuited by the null check, so an unpinned hive
never evaluates it).
16/16 `meta::` tests, clippy clean, `nix fmt` no-op, `nix build .#docs`
green.
Agents run whatever `claude-code` the meta flake's `nixpkgs` resolves
to, and that is normally a release channel. This one package moves fast
enough that stable trails unstable by weeks — 26.05 is on 2.1.187 while
unstable carries 2.1.220 — and an agent cannot fix it for itself: it
only ever sees the single nixpkgs hive-c0re injects, so an `agent.nix`
has no other tree to reach for.
New host option `services.hyperhive.c0re.claudeCodePackage` takes the
package directly and rides the existing `hyperhiveDocs` threading path —
serveConfigJson -> HiveEnv -> render_flake — to reach each agent as
`hyperhive.claudeCodePath`. Null (the default) is today's behaviour.
What travels is the store *path*, as a plain string literal, not a flake
input: containers share the host's `/nix/store`, so the build is already
reachable inside them with its whole closure and has nothing to travel.
An input would be worse than useless — a `path:/nix/store/<pkg>` input
is re-copied as a reference-less `-source`, which strips exactly the
closure the binary needs.
The catch is that a path written into a generated flake is text, so
nothing in the container's closure keeps the binary alive. The host does
that instead, and gets it for free: the package is interpolated into
`/etc/hyperhive/serve.json`, `builtins.toJSON` preserves string context,
so the /etc entry references it and the system closure gc-roots it for
as long as that generation is the one the agents were rendered from. An
assertion pins that property, because losing the context is invisible at
eval and at deploy — it would surface only as every agent failing to
spawn `claude` whenever the next gc ran.
Container side wraps the path in a symlink farm rather than putting it
on PATH directly: `systemd.services.<name>.path` and
`environment.systemPackages` both coerce a store-path *string* through
`lib.toDerivation`, i.e. `builtins.storePath`, which pure evaluation
rejects. Interpolating the path into a builder is just text and
evaluates anywhere. `claude-code` drops out of systemPackages when a
pin is set, so there is exactly one claude in the container.
Refs #2693
The hive applies one `agentCpuQuota` / `agentMemoryMax` to every
container. That's the right default and the wrong ceiling: a build-heavy
agent needs headroom the other twelve don't, and raising the hive-wide
value to suit it hands that headroom to everyone.
Adds a per-agent override, persisted host-side and resolved per-field
against the hive defaults.
Follows the existing `meta/*.json` pattern (`capabilities.json`,
`tool-groups.json`): a host-side map read by `hive-c0re`, staged and
committed in the meta repo so every change lands in the audit trail.
```json
{ "sock": { "cpu_quota": "400%", "memory_max": "8G" } }
```
Fallback is **per field**, not per agent: an entry with only
`memory_max` leaves that agent on the hive-wide CPU quota. Absent file,
absent agent and absent field all resolve to the hive default, so the
feature is inert until someone opts an agent in.
Unlike the other meta files this one is **not** injected into the
container — a limit is something done *to* an agent, not something it
reads about itself.
```
hivectl agents set-limits sock --cpu-quota 400% --memory-max 8G
hivectl agents set-limits sock --reset
```
Values are validated before they're persisted: they go into a systemd
drop-in verbatim, and a typo there makes the unit fail to *start* —
turning a fat-fingered quota into a container that won't come back.
The command is declarative: each call replaces the agent's whole entry.
That makes a forgotten flag a silent revert, so a bare `set-limits
<name>` is rejected at the clap layer and clearing needs an explicit
`--reset`.
`ContainerView` gains `cpu_quota` / `memory_max`, both always populated:
there's no "unset" state to render, only "same as everyone else". They
reflect what the drop-in *says* — what the next start will enforce — not
a live cgroup reading.
The write goes through `meta::commit_resource_limits` rather than the
bare setter, so it's staged and committed under `META_LOCK`. Writing
without committing would leave the meta working tree dirty for the next
`prepare_deploy` to trip over.
Docs: `persistence.md` (the new meta file, and why it isn't injected),
`tools/hivectl.md` (the prose guide), `tools/hivectl-cli.md`
(regenerated clap dump).
Closes: internal/requests issue 25
The two-phase approval deploy keeps a bumped `flake.lock` staged
uncommitted for the whole container build, so no other meta mutation may
land inside that span — until now enforced by a process-global
`meta::exclusive()` mutex held inside each executor fn.
A `MutexGuard` cannot outlive the fn that takes it, which is what blocks
decomposing the opaque `ApprovalDeploy` node into scheduler-visible
sub-nodes: the window has to span them. Replace the mutex with
`Resource::MetaWindow`, a global capacity-1 queue resource declared by
every meta-mutating node kind (`NodeKind::needs_meta_window`). Resources
are held by a subtree root across its whole subtree, so a later increment
can hang the deploy's phases under one window-holding parent.
Same global serialisation as before, and the scheduler now blocks a node
from being claimed rather than parking a worker on a mutex.
Split the rebuild's meta preamble out of `Prebuild` into a new `MetaSync`
node. `Prebuild` must NOT hold the window: the old mutex was deliberately
scoped to drop before the multi-minute toplevel build, which only reads
the store, and a cap-1 global held across it would serialise every
agent's rebuild behind every other's. `MetaSync` is a sibling root that
`Prebuild` deps `AfterOk` on — not its parent, since a parent's resource
covers its whole subtree and would reintroduce exactly that problem.
Queue tests: shape assertions gain the extra node, which is the point of
the change (phases become nodes). The concurrency invariants are intact
but observed one step later — the `MetaSync` heads take turns on the
window, exactly as the runtime mutex made them, so those tests now
complete the heads before asserting that the prebuilds overlap.
Per operator review (#2547): remove the never-reached branch. render_flake
read HIVE_FORGE_URL inline with an unwrap_or_else(localhost:3000) fallback
that can't be hit — the daemon always sets HIVE_FORGE_URL (environment.nix)
— and the localhost URL wouldn't match the domain-scoped credential helper
anyway. Thread forge_base as an explicit param instead: the caller
sync_agents passes forge::forge_http_base(), tests pass it explicitly. This
removes the dead branch AND the hidden env-read, and drops the racy env
set_var from the forge-url test. render_flake is pure/param-driven again.
The persistent agent-<n> input url in the meta flake is now the forge
config repo, but the deploy + manual-rebuild re-lock must not follow it:
prepare_deploy and lock_update_for_rebuild override the input back to the
local applied mirror (git+file://<applied_dir>, main head) via
--override-input, which writes the applied rev into flake.lock.
This keeps the forge as the declared, reviewable source while:
- surviving forge unreachability (rebuilds fire on crash-restart and meta
bumps too, not just config PRs -- coupling every rebuild to forge would
be a regression),
- deploying exactly the reviewed head applied/<n>/main was fast-forwarded
to (no TOCTOU on a newer forge head merging mid-deploy),
- reusing verify_commit's local-override pattern so verify and deploy eval
the same source.
New applied_override_url helper + unit test.
- deploy-window gate (meta::exclusive) + path-limited meta commits:
a perm/lock/topology commit can no longer sweep an ApprovalDeploy's
staged flake.lock and neuter abort_deploy (regression test included)
- cancel surfaces now buffer terminal roll-ups the scheduler drains,
so a queued approval DAG cancelled by the operator resolves its
approval instead of dangling, and cancelled power ops revert their
wanted flip to the observed state
- hivectl restart / restart-all ride the queue (lease serialization,
transient guard) and restart sets wanted=Up like the old kill+start
- exactly one Rebuilt event per rebuild DAG, emitted at terminal
- StopForUpdate pre-seeds a missing agent_power row from the pre-stop
observation so a rebuild can't strand an unknown agent offline
- history trim keeps terminal fan-out parents with live children
- audit_log back on db::open; swarm.js badge for reconcile DAGs
CLAUDE_CODE_OTEL_DIAG_STDERR was always set when OTEL is enabled, making
OTEL SDK errors (e.g. 404 from a misconfigured collector endpoint) appear
in every agent's stderr unconditionally. Move it behind a new opt-in flag.
Changes:
- nix/modules/hive-c0re.nix: add services.hyperhive.otel.debug (bool,
default false); wire to HYPERHIVE_OTEL_DEBUG env on hive-c0re unit.
- hive-c0re/src/meta.rs: add debug field to OtelConfig; read
HYPERHIVE_OTEL_DEBUG; emit hyperhive.otel.debug = true when set.
- nix/templates/harness-base.nix: add hyperhive.otel.debug internal
option; move CLAUDE_CODE_OTEL_DIAG_STDERR out of otelSettingsEnv into
a debug-gated lib.optionalAttrs block.
Default behaviour: OTEL exports silently (no stderr noise). Operators
troubleshooting collector connectivity set services.hyperhive.otel.debug
= true to re-enable the diagnostic output.
Add swarm.peers.<domain>.caCert (path to a peer hive's root CA PEM),
trusted everywhere the hive's own internal CA is — so a self-signed
peer hive can federate (matrix) and any in-hive consumer validates its
certs.
Mechanism (reuses the existing hive-CA embedding): the meta-flake
renderer embeds a LIST of CA files next to each agent's flake —
hive-ca.pem (the hive's own self-signed CA, when active) plus each peer
caCert as peer-ca-<N>.pem — and emits them all in
security.pki.certificateFiles, so every agent trusts them at build
time. The matrix container trusts the same peer CAs for federation TLS.
Nothing is installed in the host trust store; the certs live in the nix
store (no mutable host file).
- meta.rs: embedded_ca_files() = hive CA + peer CAs (from new
HIVE_PEER_CA_PATHS env); ca_embed_state() tracks the list (content +
add/remove); sync_agents materialises + stages the list; render emits
the multi-entry certificateFiles. Tests cover hive-only / hive+peers
/ peers-only / none.
- hive-c0re.nix: HIVE_PEER_CA_PATHS service env (colon-joined caCerts);
caCert / certFingerprint option docs updated to the hive-wide scope.
- hive-matrix.nix + docs/swarm.md: scope + comment updates.
certFingerprint stays the c0re-only leaf-pin path.
Wire agents to trust the gateway's self-signed leaf at build time. When
the gateway runs self-signed TLS, hive-tls sets HIVE_TLS_CA_PATH in
hive-c0re's service env pointing at the host hive CA cert. The meta flake
renderer reads it and, when present, writes the public CA cert next to
flake.nix as hive-ca.pem and emits security.pki.certificateFiles so every
agent's system trust store includes the hive CA.
Build-time embedding (rather than a runtime bind-mount + bundle service)
keeps trust robust: the CA travels with the flake source, lands in the
standard NixOS trust store, and needs no per-process SSL_CERT_FILE
plumbing. Only the public CA certificate is embedded; the CA private key
never leaves the host. The cert is re-embedded and re-committed on CA
rotation even when the flake is otherwise byte-identical; when self-signed
TLS is off the embedded cert is dropped so the flake stays buildable.
Covers OpenSSL-based tools (git, curl) directly. A follow-up switches the
hive-forge reqwest client to native roots so it picks up the same store.
The forwarded host env vars (HIVE_FORGE_URL, HIVE_FORGE_PUBLIC_URL,
HYPERHIVE_HIVE_DOMAIN, ...) were emitted only into the harness
service's own unit environment. Under network isolation that left the
bash-task runner, the matrix daemon, tea-login and interactive shells
without HIVE_FORGE_URL, so `hive-forge` and friends fell back to the
loopback default (localhost:3000) — unreachable from a private netns,
where the in-cluster gateway URL (forge.<domain>) is the only path.
Emit the forwarded set into systemd.globalEnvironment too, which every
unit + shell in the container inherits, so all in-container surfaces
see the same forge/matrix endpoints. The existing harness-service
forward is kept (harmless redundancy). Shared netns is unaffected (the
localhost default still works there).
Adds a regression test asserting the forwarded vars land inside the
globalEnvironment block, and the generated flake parses.
The nixpkgs bump to clippy 0.1.95 / cargo 1.95.0 added + strengthened a
large batch of lints. CI denied ALL warnings (`-D warnings`) against the
`pedantic = warn` workspace lint, so the bump hard-failed `nix flake
check` workspace-wide with zero code changes — and would recur on every
future clippy bump.
Posture fix (the durable part): CI now runs
`-D warnings -A clippy::pedantic`, so the default/correctness/style lints
stay a hard gate while the "extra, opinionated" pedantic group is
advisory only (still `warn` for local `cargo clippy` via the workspace
lints table, just non-blocking in CI). `-A` rather than `-W` so the
group drop doesn't re-enable the specific pedantic lints the workspace
allows (e.g. `must_use_candidate`).
Also fixes the genuine DEFAULT/STYLE lints the bump surfaced across the
workspace (doc_lazy_continuation, collapsible_if, ptr_arg,
match_like_matches_macro, …) via `cargo clippy --fix` + manual stragglers
(`too_many_arguments` #[allow] on the host-config constructors), and
three tests that had rotted while the CI runner was offline (#1221):
- topology::top_level_agents_in_multi_root — hardcoded unsorted expected
- rebuild_queue::depends_on_evicted_dep_counts_as_resolved — needs
MAX_HISTORY_PER_KIND newer terminals to evict, not one
- coordinator::agent_paths doctest — illustrative pseudo-code, now `ignore`
Validated: clippy + formatting + cargo-test checks all pass.
systemctl --machine=hive-gateway requires root (machine-bus transport
enters the container namespace). hive-c0re is unprivileged, so every
call to nginx_active_state() and gateway_systemctl() silently failed
with exit 1, causing a continuous 30s retry loop without ever
syncing nginx.
Fix:
- Move state-aware nginx logic into hive-priv ReloadGatewayNginx:
check ActiveState, then reload/reset-start/start accordingly.
hive-priv already runs as root and has machine-bus rights.
- Remove nginx_active_state() and gateway_systemctl() from
gateway_nginx.rs (they were always running unprivileged, always
failing silently).
- Make write(), reload_if_pending(), reload_gateway_nginx() async so
they can call the async priv_client without a blocking bridge.
- Update callers in agent_sockets::spawn_poll and meta::sync_agents
to await the now-async functions.
The priv_client::reload_gateway_nginx() call and PrivRequest::ReloadGatewayNginx
wire type already existed — the gateway_nginx module was just not using them.
- hive-sh4re: UnreadMatrix gains summary: String field (per-room breakdown)
- hive-matrix-mcp/protocol: add RoomUnread struct + UnreadSummary request
- hive-matrix-mcp/handlers: collect_unread() fetches per-room data;
single-unread rooms include truncated last-message body + sender;
multi-unread rooms carry count only
- hive-matrix-mcp/wake: format_unread_summary() builds wake body from
RoomUnread slice; terse one-liner for single-room/single-message,
bulleted list for multi-room; always appends read-hint
- hive-matrix-mcp/timeline: wake body now covers all rooms with unread
at fire time, not just the triggering event; falls back to per-event
teaser if notification counts haven't updated yet
- hive-ag3nt/mcp: matrix_unread_summary() replaces matrix_unread_rooms();
UnreadMatrix loose end carries per-room summary lines; render shows
room breakdown with sender: body for single-unread rooms
