Handle OOM kills (exit code 137) with actionable memory guidance,
filter ANSI/progress noise from error output, add --no-cache-dir to
pip installs, reduce download concurrency to 2, and bump default
container memory from 4g to 6g.
Add a "Generate strong password" button with Sparkles icon to the
Add Members form in Settings > People. Generated passwords are shown
in plain text with a copy button and an amber warning to copy before
creating the user. Also upgraded the button style on the change
password page to match. Translated copy/warning strings for all 21
locales.
Closes#139
- Bump all workspace package versions to 1.17.0
- Update APP_VERSION constant and OpenAPI spec
- Update AI tool count from 15 to 16 across docs and i18n
- Update tool table with AI Canvas Expand, Meme Generator, Beautify
- Add image editor, OIDC, and 20 languages to README features
- Add release notes for v1.17.0
- Add JSON-LD structured data and SEO improvements to landing/docs
- Filter known client-error noise (rate limit, empty body, unsupported
media type, content-length mismatch, premature close) from Sentry
via beforeSend to stop 644 events of non-actionable noise
- Sanitize x-output-filename header to prevent TypeError on non-ASCII
filenames in optimize-for-web preview (23 events)
- Handle EPIPE on Python dispatcher stdin write with graceful fallback
to per-request spawning instead of crashing (NODE-W)
- Map EACCES on storage directory/file write to proper 503 status
instead of generic 500 (NODE-P, 3 events)
ai-canvas-expand was added to constants.ts and route files but never
added to the landing page bento grid, causing all hardcoded counts
to remain at 51. This updates all references across source, docs,
i18n, and tests to reflect the correct count of 52 tools.
Add complete i18n infrastructure with 21 supported languages:
English, Simplified Chinese, Traditional Chinese, Japanese, Korean,
Spanish, French, Italian, Brazilian Portuguese, German, Dutch, Swedish,
Russian, Polish, Ukrainian, Arabic (RTL), Turkish, Hindi, Vietnamese,
Indonesian, and Thai.
- I18nProvider context with three-tier locale detection
(user preference > navigator.languages > instance default > English)
- ~1500 translation keys per locale with TypeScript-enforced completeness
- Dynamic code-splitting: only the active locale is loaded at runtime
- Language selectors in footer, login page, settings, and mobile sidebar
- Arabic RTL support with CSS logical properties across all components
- Tool names, descriptions, and categories translated via i18n helpers
- Public API endpoint GET /api/v1/config/locale for instance default
- Multi-script font stack (CJK, Arabic, Devanagari, Thai, Cyrillic)
- format() and plural() helpers for interpolation and pluralization
- API error translation mapping (translateApiError)
- 36 Playwright e2e tests verifying all 21 locales load correctly
- 25 unit tests for format, plural, locale detection, and completeness
- Updated translations.md docs and CLAUDE.md conventions
The GPU detection in gpu.py had two issues preventing GPU usage in
containers (especially rootless podman with CDI):
1. When torch was installed but torch.cuda.is_available() returned
False, the function returned immediately without trying the
ONNX Runtime + nvidia-smi fallback. This meant a CPU-only torch
build (installed before GPU was available) would block all GPU
detection, even for ONNX-based tools.
2. The failure logged a generic "torch loaded but CUDA not available"
with no diagnostic information, making it impossible to debug
whether the issue was a CPU-only build, missing libraries, or
device permissions.
The fix restructures gpu_available() into three detection tiers
(torch -> ONNX Runtime -> nvidia-smi) that always fall through on
failure. When torch CUDA fails, it now checks torch.version.cuda to
distinguish CPU-only builds from CUDA builds that can't access the
GPU, and logs LD_LIBRARY_PATH, torch.cuda.init() errors, and
nvidia-smi results.
Also fixes two env var passthrough bugs in buildMinimalEnv():
- SNAPOTTER_GPU was never passed to the Python subprocess, so the
user-facing GPU override env var had no effect
- MODELS_DIR was a dead entry (never set as env var); replaced with
MODELS_PATH which the Dockerfile sets and Python scripts read
Closes#134
Auth: login rate limit 30/min (was 500), global rate limit 1000/min (was
unlimited), password/username max lengths on all Zod schemas, session
invalidation on role change, API key legacy scan bounded to 100 keys.
SVG: hardened regex sanitizer with CDATA stripping, XML entity decoding,
set/animate/iframe/embed blocking, comprehensive data: URI blocking,
use element external href blocking. 11 attack payload fixtures added.
SSRF: fixed DNS rebinding TOCTOU by pinning resolved IPs via custom
HTTP/HTTPS agents. Added 6to4 and NAT64 to blocked IPv6 ranges.
Docker: capability dropping (cap_drop ALL + minimal cap_add), resource
limits (4g/8g mem, 512/1024 pids), healthcheck timeout, password
removed from startup banner, default password warning comments.
Network: CSP and HSTS applied in all environments (not just production),
stack traces removed from all error responses, internal paths stripped
from error details, per-route rate limits on uploads (60/min) and URL
fetches (200/hour).
Files: exclusive temp file creation (O_EXCL), disk space circuit
breaker, per-user storage quotas, settings payload 64KB size guard.
Python sidecar: script name allowlist in dispatcher, minimal environment
for subprocess spawns.
Dependencies: fixed 6 production CVEs (drizzle-orm, fastify, fast-uri,
@fastify/static, next, archiver/lodash). Pinned all GitHub Actions to
SHA hashes.
114 security tests added. Full OWASP Top 10 penetration test matrix
verified against production Docker container (30/30 pass after
hardening).
- Replace content-aware-crop with ai-canvas-expand in TOOLS[], AI_TOOL_IDS,
and FEATURE_BUNDLES (matching the already-updated tool-registry.tsx and
feature-manifest.json from commit c6a5d3f)
- Fix trailing syntax error in features.ts (extra closing brace)
- Add ai-canvas-expand-settings mock to tool-registry test files
- Update watermark-image tests to expect 400 (validation rejection) instead
of 422 (processing failure) for corrupted image buffers, matching the
actual route behavior where validateImageBuffer catches them first
- Two-gate threshold: Otsu >= 60 uses Otsu; 40-59 uses fixed 100
(catches strong scratches on borderline images)
- Remove morphological OPEN after component filtering: it was eroding
thin scratch lines that were correctly detected
- Lower Otsu gate from 60 to 40 to avoid false-negating borderline images
The transparency-fixer now directly detects the baked-in checkerboard
pattern using per-pixel chroma analysis instead of BiRefNet AI matting.
Achromatic pixels in the gray range are classified as background
(transparent), chromatic pixels as foreground (opaque), with smooth
transitions at anti-aliased edges.
- No longer requires Python sidecar or background-removal bundle
- Watermark removal uses Sharp median(5) filter pre-processing
- Moved tool from "ai" to "utilities" category
- Removed from PYTHON_SIDECAR_TOOLS and background-removal enablesTools
- Near-instant processing (pure Sharp, no model inference)
Add TIER_PARAMS dict with fast/balanced/high presets controlling band
size, mask dilation, seam strip width, and Telea pre-inpainting. Parse
tier from sys.argv[7] with balanced fallback. Conditional Telea and
seam refinement steps skip cleanly for fast tier. Progressive outpaint
now accepts band_size and progress bounds for tier-appropriate scaling.
- Fix dispatcher pipe deadlock: drain stdout pipe in a background thread
to prevent blocking when ONNX runtime output exceeds 64KB pipe buffer
- Add 5-minute SSE stall timeout so the UI shows an error instead of
hanging forever when async AI processing stalls
- Guard CPU colorization: skip for images >2MP on CPU and when DDColor
model is not installed, with clear user-facing messages
- Add AVIF decode fallback via ImageMagick for bitstream variants that
Sharp's bundled libheif cannot decode (affects all tools)
AVIF (and other Sharp-native formats) were written as raw bytes to a
.png temp file, causing PIL to fail with "cannot identify image file".
Every other AI module wrapper already converts via sharp().png().toBuffer()
before writing; face-landmarks was the only one that skipped this step.
- Refactor use-tool-processor and use-pipeline-processor hooks
- Enhance dropzone component with improved UX
- Improve seam carving with better error handling and tests
- Add JXL format encoding support to format-encoders
- Update tool routes for consistent format handling
- Add dropzone unit tests
The binary search exited early at 5% tolerance, producing output like
48.3KB for a 50KB target. Reducing to 1% with 12 iterations makes the
output much closer to the requested target.
The binary search found the right quality but sharp(buffer).toBuffer()
re-encoded at default quality 80, inflating the output (e.g. 50KB target
producing 90KB). Replaced buffer-wrapping with proper Sharp pipelines
that include .toFormat() with the proven quality. Also added progressive
dimension reduction when quality alone cannot reach the target, and
tightened tolerance to only accept at-or-below-target results.
CLAHE width/height is tile size in pixels, not tile count. A 3px tile on
a 992x1088 image created ~330x360 independent histogram regions, producing
crosshatch/etching artifacts. Now uses image_dimension/8 (clamped 8-256)
for ~8 tiles per axis. Also strips alpha before enhancement and re-joins
after to prevent CLAHE/normalise/linear from corrupting transparency.
CLAHE provides adaptive local contrast, normalise stretches the
histogram, and gamma adjusts exposure perceptually. Replaces the old
modulate/linear pipeline that compounded errors and darkened images.
Preset multipliers now include clahe and normalise entries.
Key fixes beyond the spec:
- maxSlope rounded to integer (Sharp requirement)
- White balance uses linear() instead of recomb() to avoid float-cast
that breaks CLAHE in the libvips pipeline
- CLAHE tile size adapts to image dimensions (1x1 for tiny images)
- Gamma clamped to Sharp's valid range (1.0-3.0)
- Normalise lower/upper correctly mapped to percentile cutoffs
Contrast score now uses linear stdevLum/1.2 centered at 50 (was
miscalibrated 25-75 range centered at 75). Corrections use dead zones
(score 40-60 = zero) so well-exposed images get near-zero adjustments
instead of being darkened.
