POST /video/direct/init|complete: the browser seals (SPENC2) and PUTs
straight to object storage; the api HEAD-checks, registers the unprobed
asset and queues a probe job — video bytes never cross the api box.
Worker side: produce_job_claim widened to probe jobs (042), the runner
gains a probe lane (sealed fetch → SPENC1/SPENC2 open by magic → scratch →
ffprobe → metadata write-back → complete 100), worker storage reads SPENC2
frames via DIRECT_EC_PRIVATE_KEY (inherited through compose appenv), and a
ported ffprobe util. Probe rows carrying input_json fail instead of
stranding; the claim guard no longer strands probe rows as invalid.
getObject decrypts by magic, so feeding its plaintext into openDirectHeader
always failed. getSealedObject exposes raw bytes for the header read;
DirectStorageMembers and the test fakes carry the new member.
For scale: payload bytes go browser → R2 → browser; the api handles keys,
authz and metadata only, so a cheap box can serve thousands of users.
- SPENC2 frame (browser-sealed, P-256 ECDH → HKDF wrap of a per-file CEK,
10 MiG AES-GCM segments): storage.ts gains the reader (magic-branched
getObject), presignPut/headObject, openDirectHeader; sealForDirect is the
reference sealer for cross-implementation tests
- Files API: GET /files/direct/config, POST /files/direct/init (presigned
PUT), POST /files/direct/complete (HEAD-gated commit), POST
/files/:id/direct-download (presigned GET + CEK ticket)
- web: lib/direct-storage (WebCrypto seal/open), lib/files-direct (flows),
dropzone rewritten with a multi-file sequential queue (per-file status,
direct-first with server fallback), file download decrypts in-browser for
direct objects (falls back to streaming)
- video dropzone wired to the chunked ingest API (>10 MiB → init/chunk/
complete), fixing the Cloudflare ~100 MB body cap for large videos
- DIRECT_EC_PRIVATE_KEY (P-256, PKCS8 b64) wired through compose; cross-impl
roundtrip tests prove web-seal ↔ server-open byte compatibility
Files payloads moved to R2 in 041; this removes the remaining oCIS coupling:
- delete the oCIS WebDAV client; verifyWhileStreaming moves to its own
storage-agnostic module (verify-stream.ts)
- group-files: storage_key is the only payload source; legacy rows answer
'retired storage' instead of touching oCIS
- collective share: the ocis-resource item kind is gone (peertube links stay);
share-into-library was R2-backed since the previous commit
- import-drive endpoint removed (no web caller); import-chat (homeserver) stays
- web: Drive tab + embedded oCIS page removed; Library is the Files view
Legacy e2e rows (2) deleted; every remaining archive row is R2-backed.
Cloudflare's proxy plan rejects bodies above ~100 MB, so whole-body video
uploads cannot exceed it no matter the origin limits. Clients now slice into
10 MB chunks: POST /video/upload/init, PUT /video/upload/:id/chunk/:index,
POST /video/upload/:id/complete reassembles to disk, seals, stores to R2 and
runs the normal probe + job pipeline. Max chunked upload: 100 GiB. Tests 17/17.
The Files tab was the only payload path not on object storage, violating the
2026-10-06 directive (object storage is the ONLY media I/O path) — and oCIS's
WebDAV surface is currently 500ing server-side, which made every upload fail.
Uploads now seal + dedupe into R2 via StorageService (SPENC1, content-
addressed); downloads read R2 when storage_key is set, falling back to oCIS
for legacy rows. Share-into-library copies oCIS Drive sources into R2 and
records the real sha256 instead of a placeholder. Migration 041 adds
chat_media_archive.storage_key.
The Files tab dropzone still enforced the old 20 MiB cap client- and
server-side while every other path had moved to 2 GiB. Also corrects the
stale '100 MiB' message on the video upload rejection.
The store's hand-built claim interpolated the lane predicate into the SQL
text while still passing a bind parameter — node-postgres sent 1 param for
a 0-param statement (08P01) on every poll, so the live produce worker
store-errored every ~2s and never claimed. Migration 040 replaces the
inline claim with produce_job_claim(p_lane, p_stale_after): SECURITY
DEFINER (the 035/oidc_refresh_lookup pattern — the poller is lane-global,
FORCE RLS hides every row without a tenant context), atomic FOR UPDATE
SKIP LOCKED with the queued→transcoding bump as the lease, and stale-
transcoding reclaim so a killed worker's rows return to the queue.
produce_job_workspace(p_job_id) discovers the owning tenant for the
follow-up writes, which stay RLS-bound as sp_app. pg-store rewired; store
errors now surface err.message instead of err.name; pg-store tests added.
38/38 producer tests green.
validateBrandingPatch validated storage-key values but never assigned
them into the output — PUT /workspace/branding with a logo/background
silently did nothing, so admins could never set them. Also ACCENT_HEX
gains /i (uppercase hex accepted, canonicalized to lowercase on read).
Tests unpinned from the buggy behavior; 41/41 green.
Until now collectives only existed at registration; an account with a
personal workspace had no path to start one. The creator becomes Owner
and gets a session scoped to the new workspace.
POST /video/import-drive {resourceId} pulls through the oCIS WebDAV
service-account client; POST /video/import-chat {mxcUri} pulls from the
Matrix content repository (MATRIX_MEDIA_TOKEN, homeserver falls back to
ARCHIVE_PROVISIONER_HOMESERVER_URL). Both cap at 256 MiB (buffered v1,
streaming follow-up), store through the same encrypted content-
addressed path as uploads, probe, and audit video.imported with
provenance.
WS0 of the video-producer plan: every group can open a public,
accountless upload box (random token, rotatable, disable switch).
Submitters tick the Creative Commons qualifier; credit name and a
licensing contact are optional free text — contact is visible only in
the moderation surface and never copied onto assets. Intake sniffs
magic bytes (mp4/webm/jpeg/png), caps video 200 MiB / photo 50 MiB,
rate-limits 5/hour per salted IP hash (hash never rendered), dedups by
sha256, caps 5 pending per box, and quarantines everything as pending.
Group managers approve (video → video library via the register+probe
path, photo → group asset) or reject with a note; every step audited.
Migration 035 (RLS + submission_box_lookup/submission_rate_count
definer fns). 4 unit + 4 live-DB tests.
Hub sessions gain an optional sid backed by auth_sessions (login/reissue
write the row; POST /auth/logout revokes it; pre-phase2 tokens are
refused at reissue so revocation applies from the next epoch bump).
The OIDC provider issues rotating refresh tokens (oidc_refresh_tokens,
SHA-256 at rest, family-wide revocation on reuse detection) that live
and die with the consenting Hub session — an oCIS session can no longer
outlive the sign-in that minted it, and Hub logout propagates at the
next refresh instead of the next silent-authorize. Discovery advertises
refresh_token. Migration 034; RLS via the audit_append definer pattern
for the unauthenticated token-endpoint lookup.
A boost to a gated connection (admin1/admin_majority/admin_consensus)
now holds with awaiting_publish_approval on the same publish-phase poll
the publish route uses, instead of executing the reblog. The holder
re-invokes amplify after the pass (hold-and-retry: no durable amplify
queue exists and the poll-close transaction must not do network I/O).
Unset/auto connections behave exactly as before; the route answers 202
on hold, 200 on execute. Adds amplify-governance IT suite (3 live-DB
tests).
Per the owner's spec: each social connection gets publishGovernance in its
fields — unset = manual publish (today), auto = dispatch on approval,
admin1/admin_majority/admin_consensus = a publish-phase chatgov poll
(subject draft_publish, its own open-poll slot, no collision with review
gates). Dispatch holds until the publish poll passes (E3 pre-dispatch
check); auto dispatches on approval. Includes migration 033, the PATCH
/connections/:id/publish-governance route (chatgov.manage), and 19+31 new
tests. Implementation by the branding-panel process (claude/opus), lint
and integration fixes here.
createPublishingCredentialAccessor is execution-fenced (Temporal contexts
only) — the API-process amplify route threw EXECUTION_CREDENTIAL_FENCE.
Use the established PgConnectionStore+vault accessor instead (same trust
level as the archive amplify route's bot-token path).
POST /publishing/drafts/:draftId/amplify { connectionId, kind?: repost|quote }:
the governance gate is the draft state (only approved drafts amplify), the
target is providerMetadata.sourceRemoteRef recorded at creation — never a
caller-supplied id. Connectors resolve from a process-level publishing
registry (vault-backed credentials); capability-checked per kind. Closes
the gap the governance sim documented (boosts fell back to replies).
The live tuwunel does not offer m.login.jwt (the jwt.* flat config keys were
inert), so the JWT broker flow could never work. Pivot: appservice.hub owns
the exclusive hub_* namespace on the homeserver; POST /matrix/session now
registers @hub_<username> with the as_token and mints a single-use
m.login.token via login/get_token (?user_id= impersonation). Optional
MATRIX_SSO_HS_TOKEN authenticates inbound appservice transactions, acked by
the new MatrixAppserviceController sink. Response shape unchanged.
- group-files: origin filter facet; GET /files/:id returns 410 for unavailable
rows after the ACL recheck; fix latent group_members.status filter and unbound
workspace parameter that would 500 listFiles
- archive-bot: markRedacted revokes live group_file_shares rows in-transaction
- new live-DB integration suite for the archive index surface
page.tsx now calls the real backend: GET /files with groupId/name/mimeClass/
cursor/limit (origin filter dropped — the api has no origin param), shares
tracked as ShareCreated (revoke by id), retention mapped to the orthogonal
legalHold model, downloads by item.name, samples reshaped to the wire
contract.
GroupFile gains mimeClass (the indexed coarse class the row icon and the
filter depend on): core contract, service row mapper, parseGroupFile guard,
test fixture, UI types — one vocabulary end to end. parseGroupFile's roomId
becomes null-tolerant to match direct-upload rows (mapping input stays
strict).
Verified: core build clean, api typecheck 0 errors + group-files 22/22,
web typecheck 0 errors.
Wires the three parallel group-cloud tracks to one surface and fixes the
seams between them:
- chat_media_archive.room_id becomes nullable (direct web uploads have no
Matrix room yet — the contract's null-event pattern extends to the room);
core GroupFile.roomId and ArchiveEntryInput.roomId follow.
- mime_class gains 'archive' (zip) across the 029 CHECK, validation
MIME_CLASSES and the bot classifier — one vocabulary everywhere.
- registerArchiveEntry claim keys are null-safe for room-less uploads.
- New: POST /files/groups/:groupId/upload — FileInterceptor (20 MiB) streams
through magic-byte sniffing, library-safe naming, oCIS upload and the
metadata index, behind group-membership ACL (Owner/Admin bypass).
- safeFileName/sniffMime promoted to packages/core (file-names.ts,
file-mime.ts) so the bot and the api share one implementation; the bot's
local modules become re-export shims; @types/node added to core.
Verified: api typecheck clean, group-files 22/22, bot 74/74, web typecheck
clean; 029 re-applied clean on a fresh scratch DB (room_id nullable,
archive class, claim_key unique all confirmed in the catalog).
Self-hosted multi-platform broadcast newsroom for activist collectives:
publishing pipeline with group approval and DST-proof scheduling, TweetDeck-style
column wall with hostile-content-safe rendering, live control room with Auto
Director, bridged chat governance with polls and quorum, distributed sandboxed
transcode workers, encrypted content-addressed archive, monthly analytics.
MIT licensed. See README.md and docs/.
