hyperhive/docs/tools/matrix.md
atlas e3864fe787 docs/matrix: name the [acct:<name>] prefix a multi-account agent receives
`matrix.md` documents the exact text of every inbound matrix signal —
three wake-body shapes and the invite loose-end — and none of them
mention that the daemon prefixes `[acct:<name>] ` when it serves more
than one account.

`wake::tag_account` is live on both documented paths (`timeline.rs:75`
for unread wakes, `:139` for invite todos), so an agent with an extra
account receives `[acct:ccc] [matrix] @a:s in #x: hi` where the page
promises a body starting `[matrix]`. The example is not hypothetical:
the matrix module uses `matrix-token-ccc` on dmatrix as its worked
example of an extra account.

It stayed invisible because the `None` arm returns the body unchanged,
so every single-account agent sees the documented format exactly. The
page is right for almost every reader and wrong for precisely the
readers its "Multiple accounts" section is written for.

Two placements rather than one. The prefix itself goes next to the wake
formats it corrects, with the worked example and the reason a leading
`[matrix]` match works until a second account exists. A forward pointer
goes in "Multiple accounts", because that is the section someone
configuring extra accounts actually lands on, and it previously covered
only the outbound `account` parameter — the half you pass, not the half
you parse.

Closes #4243.
2026-09-11 18:19:22 +02:00

167 lines
8 KiB
Markdown

# Matrix MCP tools and extra MCP servers
## Built-in matrix MCP (`mcp__matrix__*`)
When `hyperhive.matrix.enable = true` and the host-level matrix
tuwunel is configured, the harness autoinjects `hive-matrix-daemon`'s
streamable-http endpoint as a second MCP server (no stdio bridge —
see Architecture below). Tools land as `mcp__matrix__<name>`:
### Messaging
**Unread guard**: `send_message`, `send_dm`, `send_file`, and
`send_reply` are all rejected if the room has unread messages — call
`read_room` then `mark_read` on the latest event before sending to a
room you haven't read yet.
- `send_message(room, body)` — send a markdown message; `room`
accepts `!id:server` or `#alias:server`, daemon resolves either
- `send_dm(user_id, body)` — open (or reuse) the DM room with
`user_id` and post `body` to it
- `open_dm(user_id)` — resolve (find-or-create) the DM room with
`user_id` and return its room id **without sending anything**.
Use the returned id with room-based tools (`send_message`,
`send_file`, …) to deliver into a DM when `send_dm`'s body
parameter is inconvenient (for example for file attachments)
- `send_reply(room, event_id, body)` — threaded reply to a specific
event
- `send_reaction(room, event_id, key)` — react to a message with an
emoji key
- `send_file(room, path, caption?)` — upload a local file and post it
as a room attachment; MIME type is inferred from the file extension,
50 MiB cap; optional `caption` is sent as a follow-up text message
- `send_redact(room, event_id, reason?)` — redact (delete) a specific
event; works on your own events, redacting others' requires moderator
power level
### Reading
- `read_room(room, limit?)` — recent timeline events; media attachments
appear inline as `[file: name]`, `[image: name]`, `[audio: name]`,
or `[video: name]` markers — pass the event id to `download_file`
to retrieve the attachment
- `download_file(room, event_id, dest_path?)` — download a media
attachment from a message event to a local file and return its path;
pair with `read_room` which surfaces attachments as the markers above
- `list_rooms()` — enumerate joined rooms
(`{ id, canonical_alias, name, member_count }` per room)
- `list_room_members(room)` — members of a room
### Room membership
- `invite_user(room, user_id)` — invite `@user:server` into a room
you're already in; you must have a high enough power level.
The invitee sees a pending invite and resolves it via
`resolve_invite`.
- `resolve_invite(room, action)` — accept or reject a pending invite.
`action` is `"accept"` (join the room) or `"reject"` (decline and
leave). This is the path for invites; `join_room` is for joining a
public room you weren't invited to.
- `join_room(room)` — join a public room by id or alias. Also accepts
a pending invite if one exists, but prefer `resolve_invite` for
invites (it can reject too).
- `list_invites()` — rooms this agent has been invited to but not
yet joined (`{ id, canonical_alias, name }` per room).
### Receipts
- `mark_read(room, event_id)` — advance the read receipt
## Multiple accounts
`hyperhive.matrixAccounts` (declared in `agent.nix`) gives an agent
*additional* matrix identities beyond the hive-internal one — for example an
external-facing account alongside the internal one. Each entry is
keyed by account name and specifies `tokenFile` (bearer token,
provisioned out-of-band; basename must start with `matrix-token`),
`sessionDir` (per-account matrix-sdk sqlite state — crypto keys +
cache), and an optional `homeserver` (defaults to
`hyperhive.matrix.url`). The hive-internal account is always named
`main`, synthesized from `hyperhive.matrix.url` + agent state — this
option only declares extras, and `main` is a reserved key here.
Requires `hyperhive.matrix.enable = true`.
Every matrix tool above takes an optional `account` parameter (a name
from this map) to act as that identity instead of the primary one.
Declaring an extra account also changes what the agent *receives*: wake
bodies and invite todos gain an `[acct:<name>]` prefix — see
[Architecture](#architecture) below.
## Architecture
`hive-matrix-daemon` is a single long-running process (one per agent
container, systemd service in `nix/agent-modules/matrix.nix`) — no
stdio bridge, no separate bin. It owns the matrix-sdk `Client` + sync
loop per configured account **and** serves the matrix tool surface
directly over streamable-http on `hyperhive.mcp.matrixHttpPort`
(declared in `hyperhive.extraMcpServers.matrix` as
`{ type = "http"; url = ...; }`). Same shape as `hive-bash-daemon` and
the built-in `hyperhive` surface (`hive-mcp-http`) — claude reconnects
to the stable URL every turn instead of respawning a stdio child.
Silently exits when `<state>/matrix-token` is absent (account not yet
provisioned); the `systemd.paths.hive-matrix-daemon` watcher restarts
it the moment hive-c0re provisions the token.
Incoming room events wake the agent via `AgentRequest::Wake` with
`from: "matrix"`. The wake body format depends on the unread state:
- **Single room, one message**: `[matrix] <sender> in <room>: <body> —
use read_room to view, mark_read to clear`
- **Single room, multiple messages**: `[matrix] N unread in <room> —
use read_room to view, mark_read to clear`
- **Multiple rooms**: `[matrix] unread messages:` followed by a
bulleted list (`- <room>: <sender>: <body>` or `- <room>: N unread`
per room)
**Multi-account prefix**: when the daemon serves more than one account
(`hyperhive.matrixAccounts`), it prefixes every wake body and every invite todo
below with `[acct:<name>] `, so a wake arrives as
`[acct:ccc] [matrix] <sender> in <room>: …`. A single-account agent gets the
formats exactly as written — the daemon adds nothing — which is why matching
on a leading `[matrix]` works right up until you declare a second account.
hive-matrix-daemon includes the same per-room breakdown in the `UnreadMatrix`
entry `get_loose_ends` returns, so unread rooms surface in the
loose-ends list between turns.
**Invite wakes**: the daemon sweeps `invited_rooms()` after every sync
callback (deliberately not a one-shot `m.room.member` event handler —
a one-shot signal that raced a socket-down window was dropped with no
retry, leaving the agent deaf until manually prompted) and upserts a
todo (keyed `invite:<room>`) for each pending invite on the harness's
in-agent socket, which drives a turn. The daemon does **not**
autojoin — the agent calls `list_invites` to see pending invites and
`resolve_invite` to accept or reject them.
**Pending invites as loose ends**: the daemon upserts pending invites as
keyed todos, which appear in `get_loose_ends` output as
`[matrix] invited to <room> (<room_id>) — use list_invites to see
pending invites, resolve_invite to accept or reject`. The keyed todo
is cleared when a `resolve_invite` (or `join_room`) call resolves the
invite.
See [`docs/integrations/matrix.md`](../integrations/matrix.md) for the homeserver setup,
provisioning flow, and federation config.
## Extra MCP servers (per-agent)
Each agent's NixOS config can declare additional MCP servers via
`hyperhive.extraMcpServers.<key> = { type, command, args, env, url,
allowedTools }` — `type = "stdio"` (the default, uses `command`/`args`/
`env`) or `type = "http"` (uses `url`, a long-lived streamable-http
endpoint — see `hive-bash-daemon` and `hive-matrix-daemon` above for
the `"http"` shape). The module writes the map to
`/etc/hyperhive/extra-mcp.json`; the harness reads it at boot and
merges every entry into `--mcp-config` (under `mcpServers.<key>`)
and `--allowedTools` (as `mcp__<key>__<pattern>`).
The agent's `flake.nix` forwards every flake input to `agent.nix` as
the `flakeInputs` module arg, so you pull in external MCP-server flakes
by adding them to `inputs.*` and reference them as
`flakeInputs.<name>.packages.${pkgs.system}.default` — the resolved
sha lands in the agent's own `flake.lock` and rolls up to meta's.
`allowedTools` defaults to `["*"]`, which expands to
`mcp__<key>__*` (every tool from that server autoapproved). Restrict
to specific tool names when you want finer control.