A comprehensive role-based E2E audit (guest/host/admin, across browser
sessions) surfaced one critical and several smaller issues; this addresses
them and hardens the tests that missed them.
Critical
- The client upload pipeline was fully broken: the IndexedDB v1->v2 upgrade
opened a *new* transaction inside the upgrade callback, which throws during
a version-change transaction and aborted the whole upgrade, leaving the
queue object store uncreated -- so no UI upload ever fired. Reuse the
version-change transaction the callback provides, and bump the DB to v3 with
a contains() guard so installs already corrupted by the shipped bug
self-heal on next load. Re-enabled the previously-fixme'd UI upload E2E test.
High / Medium
- Event lock is uploads-only again: likes, comments and browsing stay open
while the event is locked (USER_JOURNEYS 9.3 / FEATURES) -- it was wrongly
freezing social interaction. Updated the event-lock spec accordingly.
- get_original now excludes soft-deleted and ban-hidden uploads, and direct
/media/originals/** serving is blocked, so a hidden user's originals can no
longer be pulled by UUID (all originals go through the checked alias).
- The upload handler reads the file field with an early-abort size cap chosen
from the declared content-type, instead of buffering the entire body before
the size check.
Low
- unban_user mirrors the ban role guard (a host can no longer unban a
host/admin banned by an admin).
- reset_user_pin's UPDATE is event-scoped.
- Admin login returns and stores a real identity (user_id + display name)
instead of a blank session.
- The host user list no longer renders target-actions (ban/promote/demote/PIN)
on the caller's own row, where the backend always rejected them.
- /diashow gains a client-side auth guard like the other protected routes.
- The join page shows the event name via a new public GET /api/v1/event.
Verified: backend cargo build clean, frontend svelte-check 0 errors, full
Playwright E2E suite 144 passed / 1 skipped.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Whole-stack review round 2: repair the broken comment button (CR1) and close
the unauthenticated export-archive leak (CR2); read-only ban model + failed-
compression cleanup + live SSE eviction (H1/H2/H3); config validation, a11y,
and hygiene (medium/low). Includes re-review follow-ups: ban guard on
edit/delete handlers, KNOWN_EVENTS registration for user-hidden/comment-deleted,
and a strengthened real-export leak test (+ truncate export-dir isolation).
Verified: backend 35 unit tests; svelte-check 0 errors; full chromium e2e
143 passed / 2 skipped / 0 failed.
The prior test asserted 404 on /media/exports/Gallery.zip against an empty
stack — it would pass even if exports were still written under /media, because
no archive was ever produced. Now it:
1. seeds an upload, releases the gallery, and polls until the real zip job
writes Gallery.zip to disk;
2. asserts the archive is NOT served from public /media (the CR2 leak); and
3. asserts it IS retrievable via the gated ticket endpoint (200 + PK zip
magic) — proving the 404 means "not public", not "no file".
Also fixes a test-isolation gap the CR2 relocation introduced: __truncate wiped
media_path but not export_path, so a real export would leave Gallery.zip on disk
and break export.spec's "ready-but-file-missing → 404" test. truncate_all now
purges export_path too. Full 06-export dir: 5/5 green, no contamination.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The round-2 re-review (full suite green) found two real defects the passing
tests missed, both now fixed and covered:
B1 — banned users could still edit/delete their OWN content: edit_upload,
delete_upload, delete_comment skipped the is_banned guard that upload/
like/comment got (round-2 removed the global extractor block and missed
these three). Added the guard (using auth.is_banned — no extra query).
The moderation ban test now asserts 403 on DELETE upload, PATCH upload,
and DELETE comment for a banned owner (+ upload survives) — it previously
only exercised POST /upload, which is why the gap shipped green.
B2 — H3 live-eviction was dead code: 'user-hidden' and 'comment-deleted' were
missing from KNOWN_EVENTS, so EventSource never listened and the handlers
never fired. Added both. New browser-driven sse-eviction test asserts a
hidden user's card actually evicts from an open feed with no reload — the
backend-only SseListener checks couldn't catch this.
Minor items folded in:
- admin dashboard bounces a mid-session-demoted admin to /feed via live
/me/context (mirrors the host page) instead of a dead error screen.
- feed "new posts" pill cleared on any full refresh (pull-to-refresh no longer
strands it).
- corrected the stale sse.rs comment (banned users may hold a read-only stream).
Verified: backend 35 unit tests; svelte-check 0 errors; e2e 04-host + comment-ui
+ export-leak 13/13 on chromium (incl. the two new regression tests).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- compression_concurrency wired to boot config (state.rs) and removed as a
dead admin control.
- patch_config gains per-key integer/range validation so numerically-valid-but-
nonsensical values (-1/NaN/3.5) are rejected instead of silently reverting to
default at read time.
- clearAuth now also clears the display name (shared-device residual).
- e2e/Caddyfile.test mirrors prod security headers + the SSE encode-exclusion so
the test stack is representative and can catch header regressions.
- .gitignore: ignore root-level Playwright artifacts (test-results/, report).
- docs: USER_JOURNEYS §10 and SECURITY-BACKLOG updated to match the implemented
read-only-ban behavior and the export-path fix.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
H1: corrected a round-1 over-reach. Bans are read-only per USER_JOURNEYS §10:
the base AuthUser extractor no longer rejects banned users (they keep read
access); Require{Host,Admin} and write handlers enforce the ban via
auth.is_banned → 403 (incl. self-unban). Demoted hosts still lose powers
immediately (live DB role) and are bounced off the dashboard. Also fixed the
login/recover redirect regression on unauthenticated 401.
H2: failed compression now self-heals instead of leaving a permanent broken
card — refund the quota, delete the orphaned original, soft-delete the row;
the frontend toasts the uploader and evicts the card on upload-error.
H3: self-delete, ban-with-hide (user-hidden), and comment-deletes now broadcast
SSE; feed, diashow, and lightbox evict the affected content live instead of
waiting for a reload. sse-eviction.spec.ts covers it.
Riders: feed/+page.svelte also carries the medium truncated-delta pill; host.rs
also carries the low atomic release_gallery claim; admin/+page.svelte also drops
the dead compression_concurrency control (medium).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
CR1: the lightbox comment button POSTed to /upload/{id}/comment (singular);
no such route exists, so every UI-submitted comment 404'd and was lost.
Fixed to /comments (plural). The prior e2e passed because its seed helper
POSTs the API directly — added comment-ui.spec.ts which drives the real
component so this can't regress silently again.
CR2: export archives (Gallery.zip / Memories.zip / HTML viewer) were written
under media_path, which is a public ServeDir — so GET /media/exports/
Gallery.zip served the entire event (every photo, caption, comment,
uploader name) to any anonymous visitor at a guessable URL, bypassing the
ticket + release gate. Moved exports to a separate EXPORT_PATH (=/exports)
on its own volume (Dockerfile chown, compose volume, gated handler reads
the new path). export-leak.spec.ts asserts /media/exports/Gallery.zip → 404.
Riders (git can't split hunks): LightboxModal also gains H3 live-eviction on
comment-deleted/upload-deleted; config.rs also wires compression_concurrency
from boot config (medium).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Re-review follow-ups (non-blocking quality items):
Tests:
- moderation H1 specs now assert exact `→ 403` instead of bare .rejects.toThrow(),
so a spurious 500 can no longer masquerade as "revocation worked".
- config: added case-insensitivity (upper/mixed-case placeholder) and the
len==32/31 boundary cases for validate_secrets.
- rate_limiter: added the trailing-comma empty-entry case for client_ip (must
fall back, not return "").
(Backend unit tests: 35 pass.)
Docs:
- README: note that with APP_ENV=production a placeholder .env makes the app
refuse to boot and Caddy wait unhealthy — reason is in `docker compose logs app`.
- SECURITY-BACKLOG + sse.rs comment: document that a mid-session ban does not
tear down an already-open SSE stream (new tickets are blocked; low blast radius).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Two items surfaced by the branch re-review:
CSP regression (blocking): script-src 'self' blocked the template-authored
anti-FOUC theme script in app.html — SvelteKit's mode:'auto' only hashes
scripts it injects. Added nonce="%sveltekit.nonce%" so it's substituted per
request and included in the CSP (survives future edits, unlike a pinned hash).
Verified: emitted CSP nonce matches the script tag; no violation, no flash.
feed_delta silent truncation: the LIMIT 200 (added earlier to kill the
stale-`since` DoS) returned only the newest slice with no signal, so a client
that missed 200+ uploads during a reconnect could not tell it should full-
refresh — the older missed uploads were dropped and unrecoverable (the next
delta advances `since` past them). DeltaResponse now carries `truncated`; the
feed's feed-delta handler calls loadFeed(true) to resync from page 1 instead
of merging a partial slice.
Verified: cargo check clean, svelte-check 0 errors, production build green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
H1: auth extractor re-reads the live user row — trusts the DB role (not the
JWT claim) and rejects banned users mid-session. e2e proves a demoted
host loses powers and a banned host is locked out and can't self-unban.
H2: client_ip takes the right-most XFF hop (the one Caddy appends); spoofed
left-most entries are ignored, restoring IP-based throttles.
H3: Caddy excludes /api/v1/stream from `encode` so SSE isn't buffered.
H4: upload — quota increment + row insert + hashtag links now one txn.
H5: feed keyset pagination tiebroken on (created_at, id) + composite index
(migration 010); feed_delta bounded with LIMIT.
H7: LightboxModal keys its comment load off upload.id, ending the
SSE-driven refetch storm.
H8: CSP via SvelteKit kit.csp (svelte.config.js) hardens the localStorage
token model against injection.
Migration 010 also adds the latent comment/comment_hashtag indexes.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
C1: reset_user_pin wrote to a non-existent column (pin_failed_attempts);
the real column is failed_pin_attempts, so every PIN reset 500'd. Fixed
the column name; new e2e (pin-reset.spec.ts) proves a reset returns a
usable PIN and the target can recover with it.
C2: config.rs::validate_secrets now rejects placeholder-ish secrets
(change_me/dev_secret/placeholder), enforces len>=32 in prod, and
requires a real ADMIN_PASSWORD_HASH. docker-compose.yml sets
APP_ENV=production so the guard actually runs. Corrected the false
"fixed" claim in SECURITY-BACKLOG.md. .env.example documents the rule.
Riders in these files (documented here since git can't split hunks):
- host.rs also carries the event-scoped ban_user fix and the
close_event/open_event no-op broadcast guard (medium).
- docker-compose.yml also adds ORIGIN (H6), app/frontend healthchecks with
Caddy waiting on health, and per-service memory limits (medium).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The two filter-search cases were test.fixme placeholders (expect(true).toBe(true)),
so the OR/AND chip rules were untested.
- Extract the grid-filter predicate from feed/+page.svelte into a pure
filterUploads(uploads, filters) in $lib/feed-filter (behavior-preserving) and
unit-test it (9 cases): tag OR, user OR, tag+user AND, AND-excludes-partial,
case-insensitive caption match, null caption.
- Replace the e2e fixmes with real tests that seed known captions/uploaders, switch
to grid view, activate chips via the search suggestions, and count grid tiles:
OR widens 1→2, AND narrows 2→1.
Frontend unit: 27 passing. filter-search e2e: 3 passing.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Follow-ups from the code review of the test-quality batches:
- Consolidate duplicated helpers into e2e/helpers/: seed.ts (seedUpload,
seedComment, listComments, findFeedRow) and sse.ts (mintSseTicket, openStream,
trackStreamOpens). Refactor authorization-deep, xss-injection, like-comment,
sse-ticket-abuse, ddos, sse-realtime, multi-tab, and SseListener to use them —
the upload/comment/ticket-flow contracts now live in one place each instead of
being re-inlined across 3–7 specs.
- xss-injection display-name loop: it navigated to /feed (which renders uploader
names, not the viewer's) so "nothing fired" passed vacuously — the payload was
never rendered. Now navigate to /account (the actual sink) and add a render
guard asserting the payload reached the DOM as escaped text before checking
__xssFired.
- sse-realtime reconnect: snapshot the stream-open count AFTER backgrounding, so
the "new connection" assertion is attributable to the foreground event and can't
be satisfied by a spurious native/error reconnect before the toggle.
- recover-page: correct the comment (auto-submit is the onPinInput handler, not an
$effect).
44 affected specs verified green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The "429 from server is surfaced (no infinite retry storm)" test routed
`**/api/v1/feed`, but the app requests `/api/v1/feed?limit=20` — a plain glob
without a trailing wildcard doesn't match a URL with a query string, so the
route never fired: `attempts` stayed 0 and `expect(attempts).toBeLessThan(15)`
passed trivially. The test never forced a 429 or exercised any retry behavior.
Fix: match with a regex `/\/api\/v1\/feed(\?|$)/` (catches the query-string URL,
excludes /feed/delta), and gate on `attempts >= 1` before judging retry
behavior so it can't pass again without actually hitting the throttled endpoint.
Found during self-review of the test-quality batches.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Stable data-testid for ContextSheet (drops animation-class coupling), aria-label
like-button locator, exact stats count, and poll-until-stable instead of a fixed
3s sleep in the 429 retry test.
- ContextSheet: add data-testid="context-sheet". longpress tests targeted the
open sheet via the `.translate-y-0` animation class (breaks on any animation
refactor) — now target `[data-testid="context-sheet"][aria-modal="true"]`,
which is stable and unambiguous vs. the centered LightboxModal (also aria-modal).
- toast-on-failure: the like button was `button.filter(hasText:/\d+/).first()`,
which could match any digit-bearing button (e.g. the comment count) → use the
stable aria-label "Gefällt mir".
- config stats: assert the exact user_count (4 = 3 seeded guests + admin) instead
of `>= 3` — deterministic after the per-test truncate, catches under/overcount.
- offline-network 429 test: replace the fixed 3s waitForTimeout with a
poll-until-the-retry-count-stabilizes (faster, and a real storm never stabilizes
→ the poll fails, which is the intended outcome).
Note: reviewed the "config restore not in try/finally" finding — it's a non-issue.
The truncate auto-fixture wipes+reseeds the whole config table (and clears the
rate limiter) before every test, so config state cannot leak between tests.
All affected specs verified green.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add a fast, DB-free unit layer: backend #[cfg(test)] modules for the rate
limiter, SSE ticket store, and hashtag boundaries; frontend Vitest (with an
$app/environment stub) for avatar and pickMediaUrl logic. 20 backend + 11
frontend tests, all green.
The suite was almost entirely e2e; pure logic had no direct coverage. Add fast,
DB-free unit tests on both sides.
Backend (cargo test — inline #[cfg(test)] modules):
- rate_limiter: allow-up-to-max-then-block, per-key independence, sliding-window
expiry, retry-after bounds, clear(), and client_ip X-Forwarded-For parsing
(first entry / whitespace-trim / fallback).
- sse_tickets: single-use consume (replay → None), unknown ticket, uniqueness +
hex shape, prune keeps fresh tickets, and an expired ticket (injected past-TTL
entry) consumes to None.
- hashtag: fill the boundary gaps the 3 existing tests missed — stop-at-non-word,
the 40-char cap (drops, doesn't truncate), no-dedup contract, and the German
umlaut truncation limitation (pinned so a future Unicode fix is deliberate).
Frontend (Vitest — new, standalone config that stubs $app/environment so
server-safe module paths import cleanly under node):
- avatar: avatarPalette (neutral for empty, deterministic, real palette entry) and
initials (?, single word, two words, whitespace collapse).
- data-mode-store: pickMediaUrl across original/saver modes and the
preview→thumbnail→original fallback chain.
- `npm run test:unit` script added.
Backend: 20 passing. Frontend: 11 passing. svelte-check: 0 errors.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Fix the SSE listener (ticket flow) and replace assert-nothing tests with real
coverage: like toggle/count semantics, comment→SSE delivery, SSE reconnect on
visibility change, per-tab SSE connection, and remove a safe-area no-op.
Several tests ran green while asserting nothing. Replace them with real
assertions, and fix the SSE helper they depend on.
sse-listener: exchange the JWT for a single-use ticket (POST /stream/ticket) and
connect via ?ticket= — the helper still used the dead ?token= scheme, so every
SSE-based assertion would have silently failed to receive events.
like-comment:
- "like is idempotent" asserted nothing (void feed; void b) → now seeds a real
upload and pins the like contract: counted once per user, toggles off on repeat
(guards double-count), and a second user's like is counted independently.
- "comment → SSE to B" asserted length >= 0 (always true) → B now subscribes to
the stream, A comments, and B must receive the new-comment event for that upload
(comment_count === 1). ~30s due to reverse-proxy SSE buffering; timeout raised.
sse-realtime: only checked a nav link was visible → now counts EventSource opens
and asserts a fresh stream connection after hidden→visible (also fixes the sim,
which set visibilityState but not document.hidden, so the close never fired).
multi-tab "SSE delivers to both": only checked nav links → now asserts each tab
opens its own stream connection (delivery isn't asserted — it hinges on the ~30s
proxy buffering; connection establishment is the reliable, honest signal).
safe-area: delete the /join probe whose only assertion was Array.isArray(x) ===
true (always true); the real sheet-level env() check already exists below it.
All verified green against the live backend.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add stored-XSS coverage for captions and comments (inert-render assertions) and
a new SSE-ticket-abuse spec (auth-required mint, single-use replay rejection,
garbage-ticket rejection).
Close two malicious-input gaps flagged in the suite review: XSS was only fuzzed
through display_name, and the SSE ticket flow had no security assertions.
xss-injection:
- Stored XSS in captions — upload with each XSS payload as the caption, mark it
feed-visible, render /feed and assert window.__xssFired stays false, no dialog
fires, and no live `img[onerror]`/`<script>` element is produced (Svelte escaping
renders it as inert text). A trailing CAPMARK gates the assertion on the caption
actually having rendered, so it can't pass vacuously.
- Stored XSS in comments — post the two render-executing payloads as a comment,
open the lightbox (which loads comments) and assert the same inert-render props.
sse-ticket-abuse (new):
- Minting a ticket requires auth (POST /stream/ticket without Bearer → 401).
- Single-use: after the first open consumes the ticket, replaying it → 401 (a 200
would be capability replay). The first open (→200) also proves a fresh ticket works.
- An unminted/garbage ticket → 401.
All verified green against the live backend.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Turn accept-both-outcomes documentation tests into pinned secure-behavior
assertions (SVG stored-XSS→400, zero-byte→400, path-traversal→201, export→404),
and replace fake-UUID authz tests with real cross-user IDOR coverage (comment
delete, upload delete, caption edit all → 403 with no state change).
Turn "accept-both-outcomes" documentation tests — which pass whether the app is
secure or vulnerable — into assertions that pin the secure behavior, and replace
fake-UUID authz tests that 404'd before ever reaching the ownership guard with
real cross-user resources.
file-upload-attacks:
- SVG-with-<script> → pin 400 (infer returns None for text → stored-XSS defense);
was [201,400], which accepted the vulnerable outcome.
- zero-byte → pin 400; path-traversal → pin 201 with UUID-derived storage and a
no-path-echo check (both were [201,400]).
- Fix the application/octet-stream rationale (no "application bypass" exists — the
handler ignores the declared type and keys off magic bytes).
authorization-deep (IDOR):
- B deleting A's comment: seed a real upload+comment as A, assert 403, assert the
comment survives, and assert the owner (A) still gets 204 — proving the 403 is
about identity, not a broken route. (Was a DELETE on the all-zeros UUID → 404
before the user_id guard, so authorization was never exercised.)
- Add B-deletes-A's-upload (403, countUploads unchanged) and B-edits-A's-caption
(403, caption intact) — the find_by_id_and_event + ownership guards.
export: pin the ZIP-download 404 (was [404,200] — a 200 is a data-exposure
regression) and split into the two real branches: not-yet-ready, and
ready-but-file-missing (new db.setExportZipReady helper).
All 21 assertions verified green against the live secure backend.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Gate always-mounted sheet dialog semantics on open state; fix focus-trap Escape
drop and deep-link back navigation; update the e2e suite for shipped magic-byte
validation, SSE ticket flow and PIN auto-submit; gate Chromium-only permissions.
The running test container had been built from an older tree; rebuilding it to
pick up current app code surfaced several tests that predated shipped
security/UX features:
- file-upload-attacks / rate-limit: assert the magic-byte rejection wording and
upload a real decodable JPEG (a zero-buffer is now rejected at the boundary).
- ddos: open SSE via the single-use /stream/ticket flow, not the dead ?token=.
- recover-page / join: the PIN field auto-submits on the 4th digit, so don't
race an explicit submit click against the ensuing navigation.
- gestures-doubletap / sheet-escape: target the lightbox by aria-labelledby and
anchor the radio accessible-name match at the start (gated dialog semantics).
playwright.config: camera/mic/clipboard are Chromium-only permissions (they threw
"Unknown permission: camera" on firefox/webkit and failed the test at context
creation); grant them per-Chromium-project. firefox-android drops the
unsupported isMobile flag.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
focus-trap: focus the trapped container synchronously on mount instead of
only after a deferred rAF. The keydown listener is node-scoped, so an Escape
pressed before the rAF moved focus into the sheet landed on an element outside
the node and was silently dropped — a real keyboard-a11y gap (open a sheet,
immediately press Escape → nothing happened). The rAF still refines focus to
the first control once laid out.
recover: replace the `window.history.length > 1` back-chevron heuristic with
SvelteKit's afterNavigate `from` signal. history.length is 2 on a fresh-tab
deep link (about:blank + page), so history.back() landed on the blank entry.
`from` is null only on a full-page load, so deep-linked users now correctly
fall back to /join (or /feed when authed).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
ContextSheet and UploadSheet stay mounted (slid off-screen via translate-y) when
closed, but declared role="dialog" + aria-modal="true" — and kept their buttons
in the a11y tree + tab order — unconditionally. So a screen reader always saw
2+ simultaneous modal dialogs, and their controls (e.g. each sheet's "Abbrechen")
collided with real open dialogs.
Surfaced by the e2e a11y suite: focus-trap asserted exactly one [role=dialog]
[aria-modal] and got 2; upload-cancel-confirm's getByRole('button',{name:
'Abbrechen'}) matched the composer's X *and* a closed sheet's button.
Fix: when closed, drop role/aria-modal and mark the subtree aria-hidden + inert
so it's out of the a11y tree and tab order entirely. Open sheets are unchanged.
Verified: chromium-mobile a11y/gesture suite now 21 passed / 5 skipped / 0 a11y
failures (focus-trap + upload-cancel-confirm green); svelte-check 0 errors.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Ports the parts of the 2026-06-27 security audit that were not already absorbed
into main:
- compression.rs: image::Limits decode cap (decompression-bomb DoS guard)
- docs/SECURITY-BACKLOG.md: triaged backlog reconciled against main
- docs/DECISION-media-auth.md: unauthenticated-UUID vs signed-gateway writeup
Verified: cargo build clean.
From the 2026-06-27 audit branch (fix/audit-2026-06-27-critical-medium), whose
work was ~80% absorbed into main via the batch branches. This ports the pieces
that were NOT in main:
- compression.rs: cap image decode with image::Limits (12000x12000, 256 MiB
max_alloc) via ImageReader instead of image::open(). The upload body-size cap
bounds the file on disk but not the *decoded* dimensions, so a small
decompression-bomb image could OOM the box during decode/resize. Surgical port
of the audit's decode guard only (not its image/video semaphore split).
- docs/SECURITY-BACKLOG.md: the audit's triaged backlog, reconciled against main
(each item tagged done / open / contingent-on-signed-media). Records the still-
open items: host moderation UI gap, owner-delete SSE broadcast, quota
mount-detection (starts_with) + low-disk guard.
- docs/DECISION-media-auth.md: writes up the one real architectural divergence —
main serves media unauthenticated (ServeDir + UUID-capability) while the audit
built a signed/TTL'd gateway. Lays out the tradeoff (leaked URL = permanent vs
~24h access; banned-user access) and a recommendation, for a human decision.
Verified: cargo build clean.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Brings the batch-2 security hardening into main (forked from the same base as
the UX batch; auto-merged cleanly — verified both sides' edits to social.rs /
main.rs coexist):
- Cross-event authorization: toggle_like / list_comments / add_comment now
event-scope the upload via Upload::find_by_id_and_event (404 on cross-event
access); delete_comment uses Comment::soft_delete_in_event. Closes the gap
where a guest could like/comment/list across events by upload UUID.
- Upload OOM backstop: the /upload route gets DefaultBodyLimit::max(576 MiB)
instead of disable(), so a multi-GB body can't be buffered before the
handler's per-class size checks run.
- upload.rs per-class size-limit refactor, XSS allowlist, deploy hardening
(Caddyfile, Dockerfiles, docker-compose, .env.example), and a data-mode
doc-comment clarifying the original-media route is capability-(UUID-)gated.
The event-scope checks sit before, and batch-3's best-effort count broadcasts
after, the like/comment mutations — both preserved.
Verified: cargo build clean, svelte-check 0 errors.
Post-commit review follow-ups for the batch-3 feed work.
Frontend — VirtualFeed: the guarded setOptions keyed only on (count, scrollMargin),
so a rotation (or the first-paint 0->real container width) changed colWidth / card
heights without invalidating the cached measurements, leaving getTotalSize and the
scrollbar stale until each row scrolled back through the window. Track appliedWidth
and call virtualizer.measure() on a width delta so heights re-estimate at the new
width (rendered rows re-measure immediately via their ResizeObserver). Covers list +
grid and the first-paint 120px-estimate case.
Backend — social.rs: the fresh like/comment count SELECT used `.await?`, so a
transient DB failure would 500 the request after the like/comment had already
committed. Make the count + broadcast best-effort (if let Ok { ... }); the mutation
succeeds regardless and the next event / pull-to-refresh reconciles the count.
Docs — FOLLOWUPS: record the review findings left as-is (grid-prepend tile reflow,
filtered-grid auto-load which is pre-existing, comment-delete stale count, and the
last-write-wins count ordering note).
Verified: svelte-check 0 errors, cargo build clean.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Address the two items from the adversarial re-review of batch-2.
- upload: remove image/heic + image/heif from ALLOWED_MEDIA. Neither the
`image` crate nor the bundled ffmpeg 6.1 (Alpine 3.21 — HEIF demuxer
only landed in ffmpeg 7.0) can decode them, so accepting them stored
posts that never got a thumbnail. iOS Safari transcodes HEIC->JPEG on
file-input selection, so this rejects only the rare HEIC-preserving
path, now with a clear error instead of a silently broken post.
- data-mode-store: correct the stale "auth-gated" comment — the
/original route is intentionally unauthenticated (UUID-as-capability)
so it works from plain <img src> / <video src>.
Re-verified: cargo build (only the pre-existing middleware.rs warning).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Address the seven findings from the security & deployment review.
High:
- upload: make infer authoritative — reject files it can't identify
(SVG/HTML/JS) and require the detected MIME to be on an ALLOWED_MEDIA
allowlist; derive the stored MIME and on-disk extension from the
detected type, ignoring client filename/Content-Type. Closes the
stored-XSS vector via media served on-origin.
- deploy: rename docker-compose.override.yml -> docker-compose.dev.yml
so the default `docker compose up -d` no longer publishes Postgres
5432 to the host; the port map is now opt-in via -f. README updated.
Medium:
- upload: DefaultBodyLimit::disable() -> max(576 MiB) as an HTTP-level
OOM backstop; handler still enforces precise per-class size limits.
- docker: run backend and frontend as non-root users.
Low:
- social/upload: event-scope toggle_like, list_comments, add_comment,
delete_comment, edit_upload, delete_upload via find_by_id_and_event /
soft_delete_in_event — cross-event IDs now resolve to 404.
- Caddy: site-wide HSTS / nosniff / X-Frame-Options / Referrer-Policy,
plus Content-Disposition: attachment on /media/originals/*.
- .env.example: replace default Postgres password with a CHANGE_ME hint.
Out of scope: localStorage JWT (root cause fixed; httpOnly cookies are a
larger change tracked separately).
Verified: cargo build (no new warnings), cargo test (3 passed),
caddy validate, docker compose config (no 5432 published by default).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Follow-up to the comprehensive code review. Five batches:
1. Cross-event authorization: host_delete_upload, unban_user, and
host_delete_comment now scope by auth.event_id. Adds
Upload::find_by_id_and_event / soft_delete_in_event and a
Comment::soft_delete_in_event variant that joins through upload.
2. Token exposure: SSE auth no longer puts the JWT in the URL.
New /api/v1/stream/ticket endpoint mints a short-lived single-use
ticket bound to the session; the EventSource passes ?ticket=...
instead. Refuse to start in APP_ENV=production with the dev JWT
sentinel; warn loudly otherwise.
3. Account hardening: per-IP+name rate limit on /recover (mitigates
targeted lockout DoS), per-IP rate limit on /admin/login, random
32-char admin recovery PIN (replaces "0000"), structured tracing
events for wrong PIN, lockout, failed admin login, ban/unban/role
change/pin-reset/host-delete.
4. DoS / correctness: comment listing paginated (LIMIT 50 + ?before=
cursor), hashtag extraction whitelisted to ASCII alnum+underscore
(≤40 chars) with unit tests, display_name / caption / comment body
length validated in chars rather than bytes.
5. Cleanup: session-touch failures now logged, DATABASE_MAX_CONNECTIONS
env var (default 10).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds a Playwright-driven end-to-end test suite under e2e/ with 134 tests
across 9 spec folders, running against an isolated docker-compose test
stack. Mobile-gesture coverage on a Pixel 7 viewport. Cross-UA smoke
matrix including a Tier-A Samsung Internet emulation. CI workflow in
.github/workflows/e2e.yml.
Also ships small fixes surfaced while writing the suite:
Backend
- JWT now carries a `jti` (per-token UUID) so two admin logins in the
same wall-clock second don't collide on session.token_hash UNIQUE.
- join handler rejects 0x00 in display names with a clean 400 (was 500
from Postgres).
- dev-only POST /api/v1/admin/__truncate route, registered only when
EVENTSNAP_TEST_MODE=1.
- rate_limiter.clear() for test isolation.
- Dockerfile: rust:1.88, COPY ./migrations into the build context.
Frontend
- Removed `aria-hidden="true"` from the leave-confirm and data-mode
backdrops on /account (it cascaded into the dialog content and made
Abmelden / Aktivieren unreachable to a11y tools).
- Bumped the PIN-copy button to ≥44×44 px touch target.
- data-testid attributes on ~20 stable interactive elements
(auth + upload routes).
Findings the suite surfaces as `[finding]` warnings on every run:
1. /admin/login has no rate-limit or lockout.
2. PIN-attempt counter races under parallel /recover requests.
3. Zero-byte uploads pass /api/v1/upload.
4. SVG-with-script can pass the magic-byte check.
Final test result: 134 passed / 0 failed / 9 skipped (test.fixme stubs
for planned gestures and one UI-upload-flow investigation).
WebKit/Firefox projects need libavif16 on the host to launch — configs
are in place; install with `sudo apt-get install libavif16` to enable.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
