First revision: DeepSeek, numeric questions, background assistant, UI fixes
All checks were successful
Verify and publish V2 container / build-and-push (push) Successful in 1m2s

Providers and pricing
- Add a DeepSeek provider: thinking mode with low/high/max effort, reasoning_content
  replayed between tool steps, cache-hit usage, /models discovery and catalog prices.
- Fix "does not support tool calling": a rejected forced tool_choice (DeepSeek thinking
  mode) now asks for the call in the prompt; only a genuine no-tools 400 is "unsupported".
- Seven reasoning-effort levels (Off to Max) mapped per adapter.
- Cached-input price per model, cache-write and DeepSeek off-peak aware cost estimates.
- Built-in bge-small embedding model (transformers.js) as the semantic search default.

Quizzes
- Numeric (type-the-answer) questions: units ignored, optional ± tolerance; editor,
  runner, importer, templates and generation prompts.
- Quiz start screen instead of showing answers; navigator ring no longer clipped and
  key legend removed; clock only for timed exams; 7-point letter grades.

Assistant and generation
- Assistant turns run on the server independently of the browser, with re-attach and
  an explicit stop; drafts tray shows generations in progress.
- Per-chat model switcher, @-tagging of class materials, full-width assistant page and
  a smaller assistant button.
- Editable generation instructions and built-in presets, separate from the chatbot
  import instructions.

Other
- Dialog footer padding, keyboard hints removed, copy buttons work over plain HTTP,
  class and folder shares include documents.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Elijah 2026-11-15 12:48:17 -07:00
parent ccdad9728b
commit f0aabe4d5d
63 changed files with 3892 additions and 488 deletions

View file

@ -103,6 +103,9 @@ function toInput(messages: ChatMessage[], caps: Partial<ModelCapabilities>, drop
return input;
}
/** The next value to try when a model rejects a reasoning effort. */
const OPENAI_EFFORT_FALLBACK: Record<string, string | null> = { none: 'minimal', minimal: 'low', low: null, medium: null, high: null, xhigh: 'high' };
interface Flags {
/** Value sent as reasoning.effort, or null to omit the field. */
effort: string | null;
@ -139,8 +142,9 @@ function buildBody(req: ChatRequest, flags: Flags): Json {
async function* stream(cfg: ProviderConfig, req: ChatRequest): AsyncGenerator<StreamEvent> {
const ctx = errorContext(cfg);
const flags: Flags = {
// "none" is only understood by recent models; if it is rejected we fall back to "low".
effort: req.reasoningEffort ?? null,
// OpenAI's top level is "xhigh". "none" and "xhigh" are only understood by recent
// models; a rejected value steps towards "medium" and is finally omitted.
effort: req.reasoningEffort ? (req.reasoningEffort === 'max' ? 'xhigh' : req.reasoningEffort) : null,
summary: true,
temperature: true,
dropReasoning: false,
@ -156,7 +160,7 @@ async function* stream(cfg: ProviderConfig, req: ChatRequest): AsyncGenerator<St
} catch (err) {
if (!(err instanceof AIError) || err.status !== 400 || attempt >= 5) throw err;
const detail = (err.providerMessage ?? '').toLowerCase();
if (flags.effort && /reasoning\.effort|reasoning_effort|effort/.test(detail)) flags.effort = flags.effort === 'none' ? 'minimal' : flags.effort === 'minimal' ? 'low' : null;
if (flags.effort && /reasoning\.effort|reasoning_effort|effort/.test(detail)) flags.effort = OPENAI_EFFORT_FALLBACK[flags.effort] ?? null;
else if (flags.summary && /summary/.test(detail)) flags.summary = false;
else if (flags.include && /include|encrypted_content/.test(detail)) flags.include = false;
else if (flags.temperature && /temperature/.test(detail)) flags.temperature = false;