a swarm o agents, each in its own nspawn cage, gossiping over unix sockets. config changes flow as git commits, the operator approves them in a browser, every deploy is a tag. cyberpunk-themed dashboard included. 💜
  • Rust 68.7%
  • Nix 15.7%
  • JavaScript 8.4%
  • CSS 3.7%
  • TypeScript 1.9%
  • Other 1.6%
Find a file
Repository files (latest commit first)
Filename Latest commit message Latest commit date
atlas 9e7a2002d1 jobq: a generic per-state roll-up, served beside the graph
A consumer that wants "how much is in flight" — a summary line, a badge,
a health check — had to fetch the whole graph and tally it client-side,
on every poll, in every consumer.

`hive_jobq_wire::state_rollup` counts `roots` and their subtrees by
state, straight off a `Graph<N, R>` with **no bound on either
parameter**. A node's state is a scheduler concept, so counting by state
needs to know nothing about what the payload or the resource are;
bounding it like the projection does would make a host implement two
display traits to be allowed to count, which is a requirement about
rendering imposed on arithmetic.

It takes the roots for the same reason `wire_snapshot` does — which
groups are in view is the host's policy, and nothing is ever removed
from a graph — so passing the same set makes the roll-up describe
exactly the graph beside it.

Each entry carries BOTH counts: `nodes` (the whole subtree) and `roots`
(just the group tops). One rebuild is ~7 nodes and 1 root, so a summary
meaning *operations* and one meaning *steps* are different numbers over
the same queue, and picking one here would make this crate decide what
counts as a job — the domain question it exists not to answer. It
reports both structural facts; the viewer chooses.

A pair, not a map: JSON object keys are strings, so a map would spell
the state twice and give the wire no ordering. Every state rides with
its zeros in a fixed order, so a consumer can index positionally and
never handles a missing bucket. Tallying positionally against
`ALL_STATES` means a new upstream `State` fails the exhaustive match in
`state_index` rather than silently landing in an existing bucket.

hive-c0re serves it at `GET /api/jobq/rollup`. The queue-side method is
a call site, not an implementation: it supplies the lock and the same
`visible_roots` as `graph_snapshot`, so the summary cannot describe a
different visible set than the graph it summarises.
2026-08-03 20:37:06 +02:00
.forgejo/workflows docs: stop claiming tracker-tag/comment-block lint are non-required 2026-07-23 00:10:59 +02:00
branding docs(#1182): remove component-diagram.svg; trim README; link to website + options 2026-06-03 19:06:06 +02:00
claude-plugins claude-subagents skill: add concurrency guidance for container memory cap 2026-08-01 11:24:58 +02:00
docs fix(#2860): no loopback default for the matrix homeserver 2026-08-03 20:34:36 +02:00
frontend swarm.js: drop the queue-summary banner rather than ship it on an interim jobq fetch 2026-08-03 20:03:49 +02:00
hive-agent always-on mark_todos_done: no ToolGroup ever exposed it, blocking every agent 2026-08-03 19:09:38 +02:00
hive-agent-mcp trim verbose todo-tool descriptions, drop exhaustive subsystem list 2026-08-02 16:41:33 +02:00
hive-agent-sock revise bulk-clear to explicit ids per mara's feedback, fix clippy line count 2026-08-02 16:41:33 +02:00
hive-bash-mcp prefix bash-task invocations with set -o pipefail 2026-08-02 16:13:36 +02:00
hive-c0re jobq: a generic per-state roll-up, served beside the graph 2026-08-03 20:37:06 +02:00
hive-core-agent-sock docs(#2627): add READMEs for the remaining infra crates 2026-07-23 13:16:29 +02:00
hive-forge hive-forge: add pr edit, reusing issue edit's shared /issues endpoint 2026-08-03 01:54:28 +02:00
hive-forge-notify feat(#2642): a github.com notification poller alongside the forge one 2026-07-31 17:23:18 +02:00
hive-host-sock hivectl: batch QueueNodes polling by id set, drop remaining dag wording 2026-08-03 20:35:24 +02:00
hive-jobq jobq: the README repeats the same false persistence claims 2026-08-03 17:28:09 +02:00
hive-jobq-wire jobq: a generic per-state roll-up, served beside the graph 2026-08-03 20:37:06 +02:00
hive-matrix-mcp fix(#2860): no loopback default for the matrix homeserver 2026-08-03 20:34:36 +02:00
hive-metric docs(#2627): add READMEs for the remaining infra crates 2026-07-23 13:16:29 +02:00
hive-priv hive-priv/hive-c0re: drop stop's SIGKILL escalation, surface a crit dashboard warning instead 2026-08-02 19:03:40 +02:00
hive-priv-sock feat(#2862): receive a passed descriptor and stream a snapshot into it 2026-07-31 22:15:37 +02:00
hive-screen-mcp docs(#2627): add README for hive-screen-mcp 2026-07-23 14:17:47 +02:00
hive-sh4re always-on mark_todos_done: no ToolGroup ever exposed it, blocking every agent 2026-08-03 19:09:38 +02:00
hive-sock-client refactor(sock): one socket client, retry as a policy value 2026-07-26 22:44:48 +02:00
hive-types docs(#2627): add READMEs for the remaining infra crates 2026-07-23 13:16:29 +02:00
hivectl job_queue: stop inventing wire data just to preserve the old source tag 2026-08-03 20:35:24 +02:00
nix wip(#2860): c0re-facing matrix.apiUrl option + HIVE_MATRIX_API_URL export 2026-08-03 20:34:36 +02:00
scripts docs: stop claiming tracker-tag/comment-block lint are non-required 2026-07-23 00:10:59 +02:00
swagger-ui-theme move swagger-ui-theme/ out of hive-c0re/ 2026-08-02 21:24:57 +02:00
.gitignore fix(review): drop libnull.rlib artifact + add Errors doc to ensure_config_pr_webhook 2026-07-11 12:19:52 +02:00
.mailmap chore(#2165): add damocles@pr1ma + lexis@pr1ma mailmap entries 2026-07-04 13:50:16 +02:00
.prettierignore docs: give turn-loop/ a README.md landing page 2026-08-03 12:55:18 +02:00
.prettierrc temp: add prettier configs 2026-07-02 23:33:11 +02:00
Cargo.lock hivectl: migrate dag_progress to hive-jobq-wire's generic GraphNode 2026-08-03 20:35:24 +02:00
Cargo.toml jobq-wire: take utoipa's chrono feature from the workspace 2026-08-03 00:39:14 +02:00
CLAUDE.md jobq: say the graph is runtime-only, because it is 2026-08-03 17:28:09 +02:00
clippy.toml hivectl: wireguard mesh setup verbs (#1756) 2026-06-19 14:37:50 +02:00
flake.lock nix flake update 2026-07-13 13:58:53 +02:00
flake.nix move swaggerUiTheme nix option out of c0re's namespace 2026-08-02 21:24:57 +02:00
README.md docs: add a top-level docs/README.md index, point README.md at it 2026-08-03 17:27:43 +02:00

hyperhive

a swarm of claude-code agents, each in its own nspawn cage, gossiping over unix sockets. config changes flow as git commits, the operator approves them in a browser, every deploy is a tag. cyberpunk-themed dashboard included. 💜

Claude code is great in one window, exponentielle across many — but only if you can keep the agents from stepping on each other, give them durable identity, and stop them from eating production. hyperhive is the substrate.

  • identity = unix socket
  • communication = sqlite-backed broker (send / recv / ask / answer / remind)
  • config = git (manager proposes, operator approves, deploys land as tagged commits)
  • blast radius = container
host (NixOS, runs hive-c0re.service)
│
├── operator
│   ├── browser → :80 (hive-gateway)    dashboard + per-agent UIs
│   │                                   /agent/<name>/ → per-agent unix socket
│   └── CLI     → /run/hyperhive/host.sock   admin protocol
│
├── hive-c0re  (Rust daemon: lifecycle / broker / approvals /
│               auto-update / dashboard / sockets)
│
├── optional containers
│   ├── hive-gateway   nginx — proxies :80 → c0re dashboard + per-agent sockets
│   ├── hive-forge     Forgejo — per-agent accounts, config mirror (agent-configs/)
│   └── hive-matrix    tuwunel — Matrix homeserver + per-agent accounts
│
└── agent containers
    ├── h-ruth     manager (privileged MCP surface, approval gating)
    └── h-<name>   sub-agent (claude + MCP tools + per-agent web UI + unix socket)

→ website · → docs · → options reference

Depth lives in docs/ (rendered at hyperhive.darkest.space/docs/) — start at docs/README.md and pick the page matching your task rather than reading front to back.

Quick start

Minimal flake.nix for a host that runs hive-c0re:

{
  inputs = {
    nixpkgs.url = "github:NixOS/nixpkgs/nixos-26.05";
    hyperhive.url = "git+https://forge.darkest.space/hyperhive/hyperhive";
    # Pin hyperhive to your own nixpkgs instead of the one it ships with
    # (see "Overriding nixpkgs" below) — recommended for most hosts:
    hyperhive.inputs.nixpkgs.follows = "nixpkgs";
  };

  outputs = { nixpkgs, hyperhive, ... }: {
    nixosConfigurations.my-host = nixpkgs.lib.nixosSystem {
      system = "x86_64-linux";
      modules = [
        hyperhive.nixosModules.default  # hive-c0re + hive-forge + hive-gateway in one import
        ({ ... }: {
          services.hyperhive.enable = true;
          # services.hyperhive.c0re.operatorPronouns = "they/them";  # default: "she/her"

          # ... rest of your host config
          system.stateVersion = "25.11";
        })
      ];
    };
  };
}

hive-c0re opens its admin socket + dashboard, auto-creates the manager container, and auto-rebuilds any container whose hyperhive rev goes stale. claude-code is unfree — hyperhive scopes the whitelist to itself, nothing for the operator to set.

Overriding nixpkgs

hyperhive pins its own nixpkgs so it builds standalone in CI. Add hyperhive.inputs.nixpkgs.follows = "nixpkgs" (as in the quick-start above) to build it against your host's nixpkgs instead — one less nixpkgs evaluation, no version drift from the rest of your system. Standard flake follows pattern; works as long as your channel is reasonably close to the nixos-26.05 hyperhive develops against. Drop it again if a much older/newer channel hits breakage hyperhive's CI doesn't catch.

For the full list of host and agent NixOS options see the options reference.

Operator CLI

hivectl is the operator-facing host CLI for ad-hoc administration that doesn't go through the broker (built alongside hive-c0re when the host module is enabled):

sudo hivectl forge create-user mara                       # provisions a forge user
sudo hivectl forge create-user mara --password 'hunter2'  # … with a fixed password
sudo hivectl matrix create-user mara                      # provisions a matrix user
sudo hivectl matrix create-user mara --password-stdin     # … reading one line from stdin

For agent names (i.e., a Coordinator::agent_state_root(name) exists), hivectl persists the resulting token to the agent's state dir like the boot sweep does. For non-agent names (e.g. the operator's own forge/matrix account), it prints the token to stdout and writes nothing.

Build / deploy

nix develop -c cargo check
nix flake check        # rust + nix + toml fmt + clippy

# deploy from a host config that imports hyperhive.nixosModules.default
nix flake update --update-input hyperhive
sudo nixos-rebuild switch --flake .#<host>