Caddy strips /api via `handle_path /api/*` before forwarding to membershipd,
so the SPA's GET /api/directory arrives as GET /directory. The route was
registered with the /api prefix, so the stripped request hit no route and
returned 404 in production: the directory never resolved and uniweb fell back
to short ids. Every other control-plane route is registered without the prefix;
this aligns directory with them.
The unit test passed despite the bug because it requested /api/directory, the
same wrong path as the registration. Corrected the request paths to /directory
so the test now exercises the real production path (verified: reverting the
registration to /api/directory now makes TestDirectoryGolden fail with 404).
Bump 0.15.0 -> 0.15.1.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add the 'Directorio de nombres (endpoint -> handle)' and 'Provisioning de bots /
unibots' sections with an end-to-end snippet, and a capability growth log entry
for v0.15.0.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add `membershipd bot add --handle <name> --out <path> [--role] [--store]` to
provision a bus identity for an automated process in a single step: mint a fresh
Ed25519+X25519 identity (cs.GenerateIdentity, the same derivation worker/chat
use), register its signing key in the allowlist, and write the credentials to a
0600 file. The file is the canonical identity format read by client.LoadIdentity,
so a worker/clientcheck binary pointed at --out connects as the new user with no
extra conversion. Shares the sqlite/kv store plumbing with `user add`.
New exported pkg/client.WriteNewIdentity writes an identity in that format but
refuses to overwrite an existing file (never silently clobber private keys).
provisionBot ordering guarantees no half-provisioned bot: refuse an existing
--out before touching the store, register (an already-registered key is a clear
error, not a panic), then write credentials. Tests cover the golden path
(register + 0600 file + LoadIdentity round-trip), default role, the
already-registered error path (no file written), and the out-exists error path
(no orphan user).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Authenticated bus users (member or admin) can now map a sender's endpoint id
back to a readable handle. The endpoint is derived server-side from each user's
sign_pub with frame.EndpointID (base64url(sha256(signPub)), unpadded), matching
the bus's own construction byte-for-byte. Only active users are listed; under
enforce the existing auth middleware rejects an unauthenticated caller with 401.
Tests cover the golden path (two users -> 200 with handles + endpoints), the
auth contract (unsigned -> 401), revoked-user exclusion, and endpoint parity
against the cross-language vector from cmd/busvectors.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Capture the reverse-proxy vhost that fronts the browser-native uniweb
client on magnus (chat-<hash>.organic-machine.com): the SPA at /, the
signed control plane under /api (prefix stripped so request signatures
verify), and the NATS-over-WebSocket data plane under /nats. One origin
means no CORS and keeps the cluster node IPs hidden behind the proxy.
Self-contained fragment (includes the shared security_headers snippet) so
it validates with `caddy validate` on its own; the other vhosts on magnus
carry basic-auth secrets and are intentionally left out of git. Documents
the matching membershipd flags this config requires (--cors-origins with
the same-origin host, --trusted-proxies naming the Caddy node).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The per-IP rate limiter keys on the transport RemoteAddr. Behind the
same-origin Caddy proxy that fronts the control plane, every request
arrives with the proxy's single IP, which collapses the limiter into one
bucket shared by the whole world — a flood from one client throttles all
of them.
Add an opt-in `--trusted-proxies` flag (comma-separated IPs/CIDRs). When
the immediate peer is one of the named proxies, clientIP now believes its
X-Forwarded-For (read right-to-left, skipping trusted hops) or X-Real-IP
and keys on the real client. A direct, non-trusted peer's forwarding
headers are ignored entirely, so this opens no quota-fanning hole: an
attacker connecting straight to the public :8470 cannot spoof a key. The
zero value (no flag) preserves the prior RemoteAddr-only behavior exactly.
Covered by ratelimit_proxy_test.go: trusted vs untrusted peers, XFF
right-to-left precedence, client-prepended forgery, X-Real-IP fallback,
and rejection of malformed proxy entries.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A browser signs every control-plane request with X-Unibus-Pub/Ts/Nonce/Sig
(busauth.signedHeaders). The CORS Allow-Headers only listed Content-Type and
Authorization, so the browser's preflight rejected the real request and the SPA
failed with 'Failed to fetch' on the first authenticated call (listRooms). Add the
four X-Unibus-* headers to Access-Control-Allow-Headers.
This was invisible to the Node smoke (fetch in Node does no CORS preflight); only a
real browser surfaced it. Verified live: enmanuel logs into uniweb against the
cluster and lists rooms. Regression test asserts the header is present.
Roll the --ws-port + --cors-origins flags (issue uniweb/0001) out to the unibus
cluster so the browser-native uniweb client can reach the data plane (nats.ws)
and the control plane (CORS) on every node. The WS reuses the data-plane TLS
(wss://) and the same origin allowlist.
Per-node WS port override (WS_PORT_<NAME>): magnus runs unibus_admin on
127.0.0.1:8480, so the bus WS binds 8485 there to avoid a crash-loop; homer and
datardos keep 8480. deploy-cluster.sh also gains DEPLOY_ONLY=<name> for rolling
one node at a time. Rolled out and verified 2026-06-13: all three nodes healthy,
WS reachable, CORS 204, cluster quorum (R3) intact throughout.
Phase 0 left the WebsocketConfig field unwired; add --ws-port so membershipd can
actually expose the browser data-plane transport. It reuses the data-plane TLS
(wss:// when TLS is on, ws:// for a loopback dev stack) and the same
--cors-origins allowlist that gates the control plane, so one flag pair opens
both planes to the browser-native uniweb client (issue uniweb/0001).
Extend the cross-language vectors with the NATS user nkey derived from the
Ed25519 public key, and a signed control-plane request (CanonicalRequest +
Ed25519 signature). These let the TypeScript busauth port verify it authenticates
on both planes exactly like the Go client (issue uniweb/0001, Phase 1).
Add cmd/busvectors, a generator that emits stable JSON test vectors for the bus
protocol and its E2E crypto (endpoint id, Ed25519 sign, ChaCha20-Poly1305 AEAD
with a fixed nonce, sealed-box of a room key, and canonical Frame wire bytes +
SigningBytes). It uses the same registry crypto (functions/cybersecurity) the
bus uses, so the vectors are the contract the TypeScript port must match
byte-for-byte (issue uniweb/0001, Phase 0).
Regenerate with: go run ./cmd/busvectors > ../uniweb/web/src/bus/testdata/vectors.json
Add Server.AllowedOrigins and an applyCORS step at the top of ServeHTTP so a
browser SPA (uniweb) can call the control plane cross-origin: an allow-listed
Origin gets the Access-Control-Allow-* headers, and a preflight (OPTIONS) is
answered 204 before the rate limiter and auth ever run. A disallowed or missing
origin gets no headers (preflight 403), so the browser blocks the request.
Wire it through membershipd's --cors-origins flag (comma list, reusing
splitRoutes as a generic parser). Empty allowlist = CORS off, no headers
emitted, behavior identical to before: native Go/Kotlin clients send no Origin
and are unaffected. Opt-in per deployment (issue uniweb/0001, Phase 0).
Tests: preflight allow/deny, header on the real response, CORS-off default, and
no-Origin native client unaffected.
Add WebsocketConfig to ServerConfig so the embedded nats-server can expose an
additional WebSocket port (nats.ws) alongside the TCP data plane. This lets a
browser SPA speak the NATS protocol directly, the way native TCP peers (Go,
Kotlin/android) already do — the first enabler for uniweb becoming a
browser-native client with no Go gateway (issue uniweb/0001, Phase 0).
The client authenticator applies to WebSocket connections too, so this adds a
transport, not a trust bypass. Plain ws:// is used only without TLS (loopback
dev); a certificate yields wss://. An empty AllowedOrigins enforces same-origin.
Nil WebsocketConfig keeps the server TCP-only, so existing single-node and
cluster deployments are unchanged.
Tests: WebSocket listener opens and completes the upgrade handshake (101); no
listener opens when WebsocketConfig is nil.
The SPA (web/) and the web gateway (cmd/webgw) move to a dedicated app
projects/message_bus/apps/uniweb (its own Gitea sub-repo). unibus is now
strictly the bus plane: membership/keys, the client library and demo peers.
uniweb consumes unibus as a Go module via replace => ../unibus.
No capability lost; same SPA and gateway, in their own service folder.
go build/vet/test green after extraction.
Add the device-local wallet onboarding to the SPA. The user's identity
is derived deterministically from a 12-word BIP39 mnemonic and lives on
the device; the browser never signs, never talks NATS, and never sends
the seed to the server.
Wallet layer (web/src/wallet/):
- derive.ts: deterministic identity from a mnemonic. seed = BIP39 seed,
then HKDF-SHA256 domain-separated into an Ed25519 signing key
(info "unibus-sign-v1") and an X25519 key-exchange key (info
"unibus-kex-v1"). The same mnemonic always yields the same sign_pub,
which is what makes recovery possible without admin intervention. The
four halves match cs.Identity on the Go side exactly.
- bip39.ts: thin wrappers over @scure/bip39 (generate, validate,
normalize) so the checksum logic stays in the audited library.
- crypto.ts: at-rest encryption of the private key with WebCrypto only
(PBKDF2-SHA256 210k iters -> AES-256-GCM). The password never leaves
the device and only protects the local key copy.
- store.ts: IndexedDB persistence of the encrypted identity (private key
encrypted; public halves + handle in the clear for display).
- account.ts: saveAndOpen / unlockAndOpen / localIdentity compose the
primitives with the gateway session API.
Screens:
- Welcome: choose invite link or recover-with-seed on an empty device.
- Join: generate seed, show it once behind an acknowledge gate, confirm
3 random words, set a local password, register the PUBLIC key with the
bus via the invite token, then open the session.
- Recover: paste the 12 words, validate, show the reconstructed sign_pub,
set a new local password, open the session. No register (the identity
is already in the allowlist).
- WalletLogin: unlock the device's stored identity with the password.
- AuthShell: shared card/header for all pre-chat screens.
- App.tsx: route between join / welcome / login / recover / chat based on
the invite link, a live gateway session, and any stored identity.
api.ts/types.ts: add register() and session() against the gateway
contract; vite dev server on :5183.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add the gateway backend for the wallet onboarding flow so each browser
session carries its OWN bus identity instead of sharing the single
operator client.
- POST /api/session (session.go): the browser hands its full wallet
keypair (unlocked from the local encrypted key, over TLS) and the
gateway spins up a dedicated bus client that acts AS that user. The
private key lives only in process memory for the life of the session
and is dropped on logout/shutdown. identityFromHex enforces the exact
key sizes (sign_pub 32, sign_priv 64, kex_pub 32, kex_priv 32) that
match cs.Identity on the Go side.
- POST /api/register (register.go): unauthenticated onboarding gated by
a one-shot invite token. Validates the two PUBLIC key halves, then
either consumes a configured --mock-tokens invite (local testing) or
proxies to the bus POST /register (--register-url, bus >= 0.12.0). The
handle/role come from the invite, never from the client.
- server.go: sessions move from a token->time map to a sessionStore of
per-user *session records; auth() now resolves the session and passes
its gateway to each handler. The legacy operator passphrase login
(POST /api/login) is kept, bound to the shared operator gateway.
- main.go: build a busTemplate config that wallet sessions clone with
their own Identity; wire --register-url / --mock-tokens.
- webgw_test.go: identity-size validation, hex-key validation, mock
token parsing, and single-use register (201 then 409) using a fixed
browser-derived wallet vector.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Replace the mock data source with a real data layer that talks to the webgw
gateway over REST + SSE. The UI components keep their look and props; only
where the data comes from changed.
- src/api.ts: the single repository layer. fetch wrappers (same-origin cookie)
for login/logout/me and rooms list/create/join/send, plus streamRoom() which
opens an EventSource and yields each decrypted message. Wire->UI mappers
(roomFromWire, messageFromWire).
- src/types.ts: add the gateway wire shapes (MeInfo, RoomWire, MsgWire) next to
the existing UI types.
- App.tsx: probe /api/me on mount to resume an existing session; otherwise show
Login. Logout calls the gateway.
- Login.tsx: the password field now unlocks the gateway session (operator
passphrase); shows a basic error and a loading state. Wallet-per-browser is
phase 2.
- ChatShell.tsx: load rooms from /api/rooms with loading / empty / error states;
same Flex layout.
- ChatPanel.tsx: stream messages over SSE for the active room (dedup by id),
composer sends through the gateway; no optimistic insert (the peer's own echo
returns over SSE with the real frame id).
- vite.config.ts: dev proxy /api (REST + SSE) -> the gateway on :8481.
mock.ts is left untouched (no longer imported) to avoid churn with the parallel
styling work on master.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add cmd/webgw: a single Go binary that holds the operator's bus identity,
connects to the bus as a real authenticated peer (pkg/client), and exposes a
small REST + SSE API the browser consumes. The browser never signs, never
speaks NATS, and never sees a private key.
Endpoints (all under /api, gated by a session cookie except login):
POST /api/login unlock a session with the operator passphrase
POST /api/logout
GET /api/me operator identity the gateway acts as
GET /api/rooms ListMyRooms
POST /api/rooms CreateRoom (default policy: encrypted+persisted+signed)
POST /api/rooms/{id}/join Join (fetch room key)
POST /api/rooms/{id}/send Publish (sealed + signed by the peer)
GET /api/rooms/{id}/stream SSE of decrypted frames (history then live)
Design notes:
- One fan-out hub per room: a single bus subscription is multiplexed to N SSE
clients, avoiding the per-(room,endpoint) durable-consumer contention that
multiple Subscribe calls would cause.
- Posture seam mirrors unibus_admin/clientcheck: empty --ca = plaintext dev,
non-empty = TLS+nkey on both planes; RefreshSession after a membership change
only under the secured (ACL) posture.
- Identity loaded from `pass` or a 0600 file, held only in memory.
- Session auth: passphrase compared in constant time; opaque HttpOnly cookie
so EventSource (which cannot set headers) can authenticate the stream.
TRUST MODEL: room content stays end-to-end encrypted on the bus. The gateway
reads plaintext only because it acts AS the operator's client — a legitimate
member of each room holding the room key. The per-browser wallet (WebCrypto)
that moves decryption into the browser is phase 2.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add a dedicated UNIBUS_NATS_MONITOR=1 toggle that opens the embedded
nats-server monitoring HTTP endpoint (127.0.0.1:8222, loopback only) so a
local metrics scraper can read /varz, /connz and /jsz for server-level
metrics (msgs/s, connections, KV bucket msgs, RAFT leader per stream,
restarts).
Previously the monitoring endpoint was only reachable via UNIBUS_NATS_DEBUG=1,
which is coupled to the verbose nats-server debug log: enabling the endpoint
also wrote routes/RAFT/room subjects to journald in clear, which regresses the
hardened posture (issue 0007). The two concerns are now decoupled.
The toggle computation is extracted to a pure function
natsLogOpts(debugEnv, monitorEnv) (noLog, debug, trace, monitor): MONITOR=1
opens the endpoint while keeping the log quiet (NoLog true / Debug false). The
inverse coupling is preserved for backward compatibility (DEBUG still implies
MONITOR). The 127.0.0.1 bind stays hardcoded — the monitoring endpoint has no
auth and must never be reachable from the network.
Deploy wiring versioned: additive systemd drop-in
membershipd-cluster.service.d/nats-monitor.conf (Environment=UNIBUS_NATS_MONITOR=1)
plus a "NATS server metrics" section in the cluster README with the rolling
activation runbook (magnus -> homer -> datardos) gated on R3 reconvergence
(followers 2/2) between nodes.
Tests: pure decoupling table (monitor on => log NOT debug; debug => monitor;
default closed) + a real embedded server with MONITOR=1 asserting /varz answers
200 on loopback:8222, and a server without the flag with the endpoint closed.
100% additive: behavior is identical without the flag. Bump app.md 0.10.0 ->
0.11.0.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add a gotcha describing the unified-storage model (the server writes
users to the same store/KV as rooms), the admin-only HTTP surface, and
the CLI-seeds-admin-#0 bootstrap. Bump the version 0.9.0 -> 0.10.0 and
add the capability growth log entry for the new HTTP admin users API.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Close the last control-plane asymmetry: rooms had a signed HTTP surface
but users were only manageable via the local CLI or direct store access.
Add admin-only HTTP endpoints, symmetric with rooms, executed against the
same privileged store the server already serves (SQLite single-node, the
replicated JetStream KV in cluster) — no new KV connection, no internal
identity, so the admin panel can manage the allowlist by signing as an
admin instead of needing --db / direct KV access.
Endpoints (all behind requireAdmin, on top of the existing
signature+nonce+TLS+enforce middleware):
- GET /users list the full allowlist (incl. revoked)
- POST /users add {sign_pub, handle, role}
- POST /users/{signpub}/revoke revoke (status flip, no hard delete)
requireAdmin is default-deny with no dev relaxation: it allows a request
only when the authenticated signer is confirmed by the store as an active
admin; any other case (no signer, non-admin, revoked, store error) is 403,
fail-closed. The request context now also carries the signer's sign_pub
hex, because the endpoint id is a one-way hash of the key and cannot be
reversed to look the signer up in the allowlist.
Validation/idempotency mirror the CLL: sign_pub must be 64-hex, role must
be admin|member (empty defaults to member), re-adding an existing key is a
409 that leaves the row untouched. The hex check is unified into
membership.ValidateSignPubHex, reused by the CLI and the handlers.
pkg/client gains ListUsers/AddUser/RevokeUser (flat UserInfo type) signed
via doJSON, so the panel plugs in directly.
Tests: non-admin -> 403 on all three endpoints; admin add->list->revoke
roundtrip; validation (400 hex, 400 role, 409 re-add, row untouched); plus
a client test against an embedded membershipd under enforce.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
New capability membershipd user add --store kv against a live cluster plus
cmd/clientcheck end-to-end verification (issue 0011 gaps, report 0012). Adds
the capability growth log entry.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Corrections learned from the real 0011 deploy:
- Bring up: the "start magnus alone and verify healthz" order deadlocks — a
lone node of a 3-node cluster has no meta-group quorum and never serves
healthz until a second node joins. Document a quorum-forming start and that
a node never self-serves.
- Replication: R1 is an unusable SPOF (all six control-plane buckets on one
node) and the cold start only converges with the three cold-start fixes;
go straight to R3 once the cluster forms.
- Add a "user add --store kv" section: the live user-add path that replaces
stop-seed-restart, with its security model and idempotency/HA/no-delete
semantics.
- Topology: real IPs, ROUTE_NETWORK=public (no WireGuard mesh exists).
- Chaos test: mark the data-plane client + failover proofs as validated (0012).
Deploy machinery now emits the persisted internal identity: the unit gains
--internal-id-file ${INTERNAL_ID_FILE} and deploy-cluster.sh writes
INTERNAL_ID_FILE into each node's cluster.env, so a fresh deploy enables the
live user-add path on every node.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The 0011 chaos test validated only the control plane (healthz + leader
failover + KV readable with 2/3); it never connected an authenticated bus
client to the data plane. cmd/clientcheck is a reusable verification tool: it
connects with a real identity (nkey + TLS on both planes, multi-node seed
lists), creates an ephemeral E2E room (encrypted + signed, no durable stream),
and either publishes N messages and asserts all come back decrypted (golden)
or publishes a counter for a duration while logging the attached node (loop),
so stopping a node mid-run shows the client fail over to a survivor and keep
receiving with quorum 2/3.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Closes the most valuable 0011 deploy gap: adding users to the running
cluster's replicated allowlist with no stop-seed-restart. Under enforce the
per-subject ACL confines every bus user to its own rooms, so no ordinary
identity may write the control-plane KV buckets; the only identity the
authenticator grants full JetStream permissions is membershipd's internal
service identity.
- main.go: --internal-id-file persists that identity (load-or-create, 0600)
instead of a fresh ephemeral key, so the same nkey is available out of
process. Empty keeps the ephemeral default (single-node/dev unchanged).
- users_kv.go: connectKVStore loads the persisted identity, presents its
nkey (recognized as internal -> full perms), opens the KV store and
writes. Defaults assume an on-node loopback invocation; a remote target
without --ca is refused (allowlist must not travel cleartext, audit N6).
Prints KV_UNIBUS_users replication (followers_current) after a write.
- users_cli.go: --store kv on add/list/revoke. Re-adding a key is an explicit
ErrUserExists (no silent overwrite / role flip); revoke is a status flip.
- pkg/client: LoadIdentity (load-only) extracted from LoadOrCreateIdentity,
preserving its "corrupt file is an error, not silently regenerated" guard.
- kv_useradd_test.go: golden write under enforce, idempotency, unreachable
endpoint, and remote-without-CA refusal against an embedded node.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Set magnus's public IP (135.125.201.30) and switch ROUTE_NETWORK to "public":
the three nodes have no WireGuard mesh (homer/datardos do not even have wg
installed), so server-to-server routes go over the public IPs, still protected
by the separate cluster route CA (mutual TLS). KV_REPLICAS is raised to 3 now
that the cluster runs at R3.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Bringing up the 3-node cluster from clean stores never converged: every node
looped on `open KV bucket "UNIBUS_rooms" (replicas=1): context deadline exceeded`.
Three independent defects in the clustered bootstrap path, none of which surface
on a single node (where JetStream is ready instantly), caused it:
1. embeddednats: route connection pooling (nats-server 2.10 default pool of 3)
churned with "duplicate route"/"client closed" reconnects on the small cluster,
interrupting the meta-group RAFT heartbeats and forcing perpetual leader
re-elections. Set Cluster.PoolSize = -1 (single route per peer).
2. embeddednats: the cluster nodes are Docker hosts, so NATS advertised the docker
bridge IPs (172.x / 10.0.x) to peers, which then tried to dial those private,
mutually-unreachable addresses. Set Cluster.NoAdvertise = true so only the
explicit public-IP routes are used. Also added a UNIBUS_NATS_DEBUG env toggle
(off by default) that enables the embedded server's logger and loopback
monitoring port for debugging the route/meta layer.
3. membership.OpenJetStream: a KV op is a NATS request/reply; on a cold cluster the
op was published once, before the node had contact with the meta leader, so the
request was dropped and the single long-context call just blocked until timeout.
Retry each bucket op with short per-attempt contexts until it succeeds or an
overall bootstrap budget (120s) is exhausted, so it lands once the meta settles.
With these the cluster forms cleanly, creates the KV buckets, scales R1->R3 in
place, and survives loss of one node (quorum 2/3). Verified on magnus+homer+datardos.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
SPA React 19 + Vite + Mantine v9 en modo oscuro (acento índigo), datos mock para
iterar el diseño antes de cablear el gateway. Login con identidad + contraseña
(la contraseña desbloqueará la identidad Ed25519 cifrada en el dispositivo).
Sidebar: avatar de usuario, buscador (rooms/usuarios/mensajes) y lista de rooms
con candado E2E / hash cleartext / badges de no leídos. Panel de chat estilo
Element (avatar+nombre+hora+texto) con composer interactivo.
Limpieza de los frontends de prueba (SPA React, app Kotlin, gateway playground,
binding gomobile) tras la fase de exploración. El bus (cmd/membershipd + pkg/*)
queda intacto y verde. Empezamos un frontend web nuevo desde cero, construido
de forma incremental. Todo lo borrado permanece en el historial git por si hay
que recuperar algo.
All seven phases (0006a–0006g) merged: blockers N3 (replicated nonce) and N2
($JS.API.> KV leak) closed, decentralized KV store wired (--store kv), homogeneous
cluster posture enforced (N1), RefreshSession in all clients (N4), the lows
(secret out of argv, migrate guard, R1/CA docs), and the 3-node deploy material.
Full suite + every audit-0008 attack regression green; govulncheck 0 reachable.
See report 0009.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Parameterized, NO-VPS-touched material to bring up unibus as a 3-node cluster.
The authoring agent ran none of it on a host; every remote-changing step is
marked HUMAN and deploy-cluster.sh defaults to a dry run.
deploy/cluster/:
- nodes.env — topology (cluster name, ports, per-node rows). Public IPs known
(homer 141.94.69.66, datardos 51.91.100.142) pre-filled; magnus public IP and
all WireGuard IPs are <PLACEHOLDER> for the human; scripts refuse to run while
any remain.
- generate-cluster-certs.sh — mints a SEPARATE cluster route CA + a route cert per
node (server+clientAuth, mutual routes) and a data-plane server cert per node
signed by the reused client CA (../tls/ca.*); SAN = public + WG + hostname.
- membershipd-cluster.service — one unit, parameterized per node via
/opt/unibus/cluster.env: enforce + per-subject ACL + TLS + --store kv,
--cluster-pass-file (secret out of argv), Restart=always.
- deploy-cluster.sh — cross-build linux/amd64, generate each node's cluster.env
(routes to the other two on the WG mesh, no userinfo), rsync + install (only
with --yes); staggered start is manual.
- README.md — runbook: prerequisites, loopback bootstrap to seed the first admin
into the KV (works around the user-CLI/KV chicken-and-egg), staggered bring-up,
verify posture+quorum, scale R1->R3 in place, and the chaos test (left to 0003f
on the real VPS).
- .gitignore — out/, build/, secrets/, *.key never committed.
bash -n passes on both scripts; go build/test unchanged.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Low-severity cluster hardening from audit 0008:
- Route secret out of argv (N1-low): --cluster-pass and a nats://user:pass@host in
--routes are visible in ps/journald. New --cluster-pass-file and the
UNIBUS_CLUSTER_PASS env var (precedence file > env > flag); the resolved secret
guards the route layer and is injected into bare --routes entries
(injectRouteCreds), so peers can be listed as nats://host:6250 with no secret in
argv. The legacy --cluster-pass stays for dev/compat.
- migrate-to-kv confidentiality (N6): refuse a remote --nats-url without --ca (the
allowlist would travel cleartext); loopback targets are exempt (isLoopbackURL).
- Docs (N1 route CA, N3 DoS): deploy/README gains a Clustering section — use a
SEPARATE cluster CA for routes (not the client CA), keep the secret out of argv,
run migrate-to-kv loopback/TLS only, and R1 is a SPOF of auth (not HA); R3
quorum is real HA. The generated cert material lives in deploy/cluster/ (0006g).
Tests:
- TestResolveClusterPass (file > env > flag precedence; missing file errors),
- TestInjectRouteCreds (injects only into userinfo-less routes; preserves overrides),
- TestIsLoopbackURL (loopback vs remote vs malformed).
CGO_ENABLED=0 go build/vet/test green; govulncheck 0 reachable.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A secured bus freezes per-subject permissions at connect time, so a peer that
creates or joins a room after connecting cannot pub/sub on it until it reconnects
(RefreshSession). No client called it, so under enforce+ACL the demos failed
closed — pushing the operator to disable the ACL (a security regression at the
operator's discretion).
Wire the membership-change contract into every client:
- cmd/worker: RefreshSession after CreateRoom, before publishing.
- cmd/chat (simple): RefreshSession after CreateRoom+Join, before Subscribe.
- cmd/chat (encrypted demo): A refreshes after CreateRoom; B refreshes after the
invite+join, both before pub/sub.
- local_files/bridge (gateway): RefreshSession after CreateRoom+Join, before Subscribe.
- mobile: new Session.RefreshSession wrapper + the contract documented for callers.
Contract (documented on the wrappers): after ANY membership change, call
RefreshSession BEFORE pub/sub on the new room (it drops active subs, so it must
precede Subscribe). On an unsecured/dev bus it is a harmless reconnect.
Test:
- TestClientCreateRoomRefreshPublishFlow: end-to-end under enforce+ACL, a peer
creates a room, refreshes, invites a second peer who joins+refreshes+subscribes,
and the publish is received — no manual intervention, the ACL stays on.
CGO_ENABLED=0 go build/vet/test green; govulncheck 0 reachable.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
egutierrez
agent