* Use Ollama /api/embed endpoint for vector embeddings
The deprecated /api/embeddings endpoint does not properly support the
truncate parameter, causing "input length exceeds context length" errors
when vectorizing files. Migrate to /api/embed which correctly handles
truncation and supports native batch input.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Wrap single Ollama vector calculation into batch
Fixes https://github.com/SillyTavern/SillyTavern/pull/5221/changes#r2850052729
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Cohee <18619528+Cohee1207@users.noreply.github.com>
* backend, frontend, bugfixes
* Mobile button and sizing
* lint
* clear folder thumbnailFile on delete, rename thumbnailFile on rename
* use filteredImages when changing sort option
* Address all the review comments
* Fix friendly title generation to handle empty strings
* Move add folder button to the header
* instead of search filtering the backgrounds in a folder and showing the folder if the results > 0, search the folder names.
* Trade button places
* Adjust button text
* feat: restrict folder creation to the Global tab
---------
Co-authored-by: Cohee <18619528+Cohee1207@users.noreply.github.com>
* Added filter for OpenRouter models provider selection
Now if a model is selected, only available providers for that model will show. Wanted to do the same for the quants, but I think the API is not returning the quants available for each model at the moment. Used existing API that for some reason was not consumed.
* Added filter for OpenRouter providers
Now if a model is selected, only the providers available show. Wanted to do the same with the quants but it seems the OpenRouter API is not giving the available quants list at the moment for each model.
* gua
* Now it also works on chat completion and only disables options
* detail
* Warning added
* eslint
* Move inline styles to CSS
---------
Co-authored-by: Cohee <18619528+Cohee1207@users.noreply.github.com>
* feat(openrouter): disable reasoning if "Request model reasoning" is disabled
* feat(openrouter): map minimum reasoning to none if request reasoning is off
* Add hint how to disable reasoning
---------
Co-authored-by: Cohee <18619528+Cohee1207@users.noreply.github.com>
* Initial plan
* feat: preserve image width/height when generating swipe images
Save width and height in ImageGenerationAttachmentProps when creating
MediaAttachment objects, and apply saved dimensions when generating
image swipes. Falls back to current extension settings if not specified.
Co-authored-by: Cohee1207 <18619528+Cohee1207@users.noreply.github.com>
* fix: validate dimension override with Number.isInteger and skip setTypeSpecificDimensions when override is present
Co-authored-by: Cohee1207 <18619528+Cohee1207@users.noreply.github.com>
* fix: call restoreOriginalDimensions when setTypeSpecificDimensions was used
Co-authored-by: Cohee1207 <18619528+Cohee1207@users.noreply.github.com>
* Contain the logic to setTypeSpecificDimensions
* Diff clean-up
* feat: enhance prompt refinement with negative input and saved resolution options
* fix: Only save image dimensions when running from command
* Fix wording in refine prompt dialog
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: Cohee1207 <18619528+Cohee1207@users.noreply.github.com>
* fix(openrouter): forward reasoning across active tool-call chains
* feat(reasoning): add tool-chain forwarding toggle and honor edited reasoning
* feat(reasoning): add OpenRouter interleaved forwarding modes
* moved the reasoning forwarding dropdown into a separate line
* feat(reasoning): default tool reasoning forwarding to disabled
* refactor(openrouter): move tool reasoning mode to CC settings
Move OpenRouter tool reasoning forwarding control to response configuration and scope it to OpenRouter.
Store mode in chat completion settings (presettable), remove legacy power_user boolean/fallback, and use constants for mode values.
Preserve OpenRouter Gemini signature forwarding independently from plaintext tool reasoning mode.
* fix(openrouter): tighten active-chain reasoning forwarding
Use trailing contiguous tool-chain boundary for active-chain eligibility.
Also rename the UI control to Interleaved Thinking Forwarding and place selector on its own line.
* fix(openrouter): use adjacent assistant reasoning for tool calls
For interleaved thinking forwarding, source reasoning only from the immediately preceding assistant non-tool message.
Keep mode gating behavior unchanged and avoid history-window reasoning carryover.
* fix(openrouter): skip tool messages for reasoning source
When forwarding interleaved reasoning, ignore intervening tool result messages when resolving the preceding assistant reasoning source.
This keeps only the first tool call in a chain tied to a prior assistant reasoning block unless a later invocation carries its own reasoning.
* fix(openrouter): keep plaintext reasoning with signatures
Do not suppress forwarded tool-call reasoning when thought signatures are present.
* fix(openrouter): split interleaved thinking mode behavior
Restore distinct mode semantics: active_chain uses nearest assistant-text boundary after skipping tool/tool-call messages, while since_last_user scans for latest assistant reasoning since user.
Update UI label to Interleaved Thinking with right-aligned dropdown and explanatory tooltip.
* style(openrouter): align interleaved thinking dropdown row
Match OpenRouter interleaved thinking control layout with existing oneline-dropdown patterns.
Also update reasoning-forwarding inline comment wording for current mode behavior.
* docs(ui): clarify interleaved thinking tooltip
Use explicit API-request wording for OpenRouter interleaved thinking tooltip text.
* i18n(openrouter): localize interleaved thinking UI
Add locale keys for OpenRouter interleaved thinking label, mode options, and inline helper description.
Wire dropdown option text to data-i18n in index.html.
* fixed helper text wrapping
* fix(ui): make interleaved thinking helper text wrap
* i18n(openrouter): translate interleaved thinking labels
Replace placeholder English values for interleaved thinking keys in non-English locale files.
* fix(ui): restore interleaved thinking dropdown alignment
* Remove changes from en.json
* Type fixes
* Reworked the interleaved reasoning provider logic
* Renamed the variables in preparation for potential implementation for other providers
* Gate interleaved tool reasoning on reasoning request setting
---------
Co-authored-by: Cohee <18619528+Cohee1207@users.noreply.github.com>
* build llama.cpp embedding url path by appending instead of overwriting
When using a llama-swap frontend, this allows choosing the embedding model
by path; for instance, with the secondary embedding endpoint URL configured
to http://127.0.0.1:5001/upstream/emb-model to pick the "emb-model" model.
With this approach, the same llama-swap instance can be used to serve both
the main LLM and the embedding model from local llama-server commands.
* Trim trailing v1 from URL
---------
Co-authored-by: Cohee <18619528+Cohee1207@users.noreply.github.com>
* check png for apng headers
* isAnimated flag
* refactor: centralize background animation extension checks
* refactor: scope animated extension dedupe to backgrounds
* remove precompute from startup
* Fix animation preference not being loaded, fix type-check of customInputs
* Fix eslint
* Fix sort on removal type
* Update metadata before returning from endpoint on CRUD operations
* Remove race condition in metadata load
* Load metadata after loading the list
---------
Co-authored-by: Cohee <18619528+Cohee1207@users.noreply.github.com>
* Add gallery argument to /imagine to skip character gallery save
Adds a `gallery` named argument (default: true) to the /imagine slash
command. When `gallery=false`, the generated image is saved to the
generic images directory instead of the character-specific gallery
folder, while still returning a usable image URL.
This is useful for extensions that manage their own image display
(e.g. inline in chat messages) and don't need gallery copies.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Can't assign to a constant
* Update help string to clarify image saving behavior in generation requests
* Fix file name template if no char name provided
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Cohee <18619528+Cohee1207@users.noreply.github.com>
- Add 'textarea' as a valid input type alongside 'checkbox' and 'text'
- Add optional `rows` property to control textarea height (default: 1)
- Implement textarea rendering with proper label, placeholder, and styling
- Update result collection logic to handle textarea values like text inputs
- Update warning message to include textarea in supported types
* feat: add prefers-reduced-motion and prefers-contrast media queries
- Integrate OS prefers-reduced-motion into existing JS toggle system:
when OS preference is active, force reduced_motion on and disable
the UI checkbox with an explanatory tooltip
- Add body.reduced-motion CSS rules for hardcoded animations that
bypass the --animation-duration variable (text_segment fade-in,
dragover pulse)
- Add prefers-contrast: more media query for enhanced focus outlines
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* Support translating title text
* Use duration variable
* Removed stream fade-in override
* Combine hi-contrast selectors
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Cohee <18619528+Cohee1207@users.noreply.github.com>
* Initial plan
* Overhaul /bgcol command: Oklab color space, dominant color extraction, full theme palette generation
- Create ThemeGenerator module in public/scripts/util/ with:
- Oklab color space conversions (sRGB ↔ Oklab ↔ LCH)
- Chroma-weighted dominant color extraction (vivid over muddy averages)
- Color theory palette: complementary, analogous, triadic hue relationships
- WCAG contrast ratio enforcement (≥3.5:1 for all text colors)
- Replace old setAvgBG with new implementation using ThemeGenerator
- Save generated theme as "Generated - <background name>" instead of overwriting
- Update command help string to reflect production-ready status
Co-authored-by: Cohee1207 <18619528+Cohee1207@users.noreply.github.com>
* Extract hue shift constants for clarity (code review feedback)
Co-authored-by: Cohee1207 <18619528+Cohee1207@users.noreply.github.com>
* Add name and bg arguments to /bgcol command
- Add `name` named argument to override the generated theme name
- Add `bg` named argument with enum provider listing available backgrounds
from /api/backgrounds/all to override the current background
Co-authored-by: Cohee1207 <18619528+Cohee1207@users.noreply.github.com>
* Improve types, add enum provider, reduce transparency
* Adjust alpha channel for better readability
* Refactor ThemeGenerator to work directly in Oklch, removing intermediate Oklab conversions
Replaced srgbToOklab/oklabToSrgb + oklabToLCH/lchToOklab with direct
srgbToOklch/oklchToSrgb functions. All color manipulations now use
{L, C, h} directly without converting through {a, b} intermediates.
The ensureContrast function now takes (L, C, h) parameters instead of
an Oklab object.
Co-authored-by: Cohee1207 <18619528+Cohee1207@users.noreply.github.com>
* Add custom background URL check and enhance theme generation logic
---------
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: Cohee1207 <18619528+Cohee1207@users.noreply.github.com>
* Fix: HTTP Basic Auth fails when password contains colons
The credentials in HTTP Basic Auth are formatted as base64(username:password).
Per RFC 7617, the username must not contain a colon, but the password can.
The previous code used `.split(':')` which splits on all colons, truncating
passwords that contain ':' characters. Fix by splitting only on the first colon.
* Use spread syntax for credential parsing
Cohee