fcd4eead56e9779b432eddc5fd46f9d2a48af0b5
4
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Bohdan Triapitsyn
|
1d17cb87b3 |
feat(walkthrough): write walkthroughs in the reader's language
A guided explanation is only useful in a language the reader reads, so the panel header gets a language picker alongside the model one, defaulting to the interface language. Like the model, it is request state rather than a setting: the language travels with the read and the generation, and the one a walkthrough was written in is stored with it, so reopening a review describes what is there instead of what a fresh one would be. Only prose is translated. Hunk aliases resolve back to hunk ids and icon/importance are validated against fixed English values, so a translated one would be dropped by the normalizer — silently losing an anchor or a style. Identifiers and paths stay as they appear in the code. The language is part of the cache key, and a read now asks the cache for the exact request it was given before falling back to the pointer. Without that the panel answered a request to switch languages with the text it already had, leaving the other language unused in the cache. Alongside it: - The answer budget is derived from the resolved model instead of a flat 24k. That number was the same for a 64k-context model and for one that admits to 384k output tokens, and on the latter it was the only reason generation failed: the model spent the whole allowance reasoning and returned nothing. It is now min(96k, max(24k, a quarter of the context)) capped by the catalog's output limit, decided once so the input reserve and the request cannot drift apart. - A read no longer offers Cancel. It is a few hundred milliseconds of git with nothing to cancel, and the button flickered on every model or language change. When the panel is showing a fallback, a banner names what is on screen versus what was asked for — only once the read has settled. - The header keeps one 32px control height and drops its labels below 680px instead of squeezing them to two letters and an ellipsis. Docs and module documentation updated in every locale. |
||
|
Bohdan Triapitsyn
|
34d0ff7383 |
feat(walkthrough): guided AI walkthrough for diffs, branches, and PRs (#2572)
A diff is ordered by file path, which is almost never the order in which a change makes sense. This adds a Walkthrough surface that reorders it: the model groups related hunks into stops, explains what each group changes about behavior, and orders the stops so each builds on the last. It explains and orders; judging code stays with the existing Review action. Reviews uncommitted work (all, staged, unstaged), a branch against its base, or a pull request. Generation is always user-initiated — nothing runs on a timer, on a file change, or as a side effect of opening a panel. Invariants worth preserving: - Hunk identity is derived on the server and only there. Ids are content hashes, so an anchor that no longer resolves is proof the code it described changed, and staleness needs no heuristics. The client matches ids to ids and never recomputes them; two implementations would have to agree forever. - The digest is never truncated. A diff that does not fit the model's context is refused with an actionable reason, because a walkthrough written against half a diff reads as confident and is wrong. - Nothing disappears. Lockfiles and other generated output are excluded from the model's input by name — never by size — and everything no stop covers is listed at the end, so "have I seen all of it" stays answerable. - Cost is explicit. Results are content-addressed, so returning the working tree to an earlier state costs nothing; generation outlives its request, so a refresh detaches the client rather than discarding paid-for work, and only an explicit cancel stops it. Supporting changes to shared modules: - git: expose the existing getRangeDiff as GET /api/git listUntrackedPaths and getUntrackedDiffs. The latter resolve the repository once for a batch instead of per file, taking a panel ~340ms on an 80-file working tree. - small-model: structured output across four wire forma and abort signal, and an onOverflow policy so an oversized prompt fails loudly instead of being silently clipped. A provider remembered so the prompt-side fallback goes first next time. - models.dev metadata: surface structured_output as tri false blocks a model, a missing field does not, because the catalog omits it for roughly half of all models. Desktop and tablet only: VS Code serves Git through its these routes, and the mobile shell does not consume the surface registry. Docs: packages/docs walkthrough page in English and all eight locales. |
||
|
Andrey Meshkov
|
7e248d4e9b |
Fix: small model dispatch fails for custom OpenAI-compatible providers (#2134) (#2135)
* fix: add support for custom provider base URLs from config * feat: read custom provider apiKey from config and use it as primary credential when it exists * doc: update small-model documentation |
||
|
Bohdan Triapitsyn
|
28f0736d69 |
feat: small-model utility calls on existing OpenCode providers (#2049)
Adds a server-side "small model" capability: direct, cheap LLM calls that reuse the user's existing OpenCode provider logins — the mechanism OpenCode uses internally for titles and summaries but does not expose through the SDK or plugins. Zero new dependencies; plain fetch with per-provider wire formats, credentials never leave the server. Core (packages/web/server/lib/small-model): - Resolution mirrors OpenCode's session scoping: explicit settings override → small_model from the OpenCode config → family scan within the session's provider → the session's own model. The global provider scan only serves callers without a session context, and background callers forbid it entirely (restrictToPreferredProvider), so conversation content never reaches a provider the user didn't pick — explicit choices excepted. - Per-provider auth replicating OpenCode's plugin loaders: GitHub Copilot (device token as bearer, no exchange), ChatGPT plan via the codex Responses API (single-flight OAuth refresh written back to auth.json), Anthropic messages, Google generateContent, generic OpenAI-compatible. - OpenCode's free models (opencode/big-pickle, *-free) are never called directly; unauthenticated providers are skipped by design. - Prompt clamping to the model's catalog context limit; thinking disabled where a wire switch exists (Z.AI/GLM, MiniMax-M3, Gemini Flash); robust content parsing with a clear error when a thinking model spends its whol budget on reasoning. - Settings → Sessions gains a Small Model group: use-default checkbox plus an override picker limited to authenticated providers, persisted with web/desktop/VS Code sanitization parity. Consumers: - Session assist: a server-side watcher on the global SSE hub generates a short recap and one suggested follow-up after a session idles quietly fo a minute, stored on session metadata (openchamber.assist). Freshness is keyed to the last assistant message id, so new activity invalidates the payload everywhere with no extra writes. The chat shows the recap under the last message after five quiet minutes and the suggestion as a dismissible chip above the composer (tap fills the input, never sends). Gated by a new Chat setting (default on) that is a hard generation switch. Language is anchored to the conversation itself, with a script-mismatch guard against model/backend language hallucination. - TTS: a third input mode, summarized — long replies are condensed to spoken prose before playback on any TTS engine. - Git: commit-message and PR generation moved off the active chat session onto the small model fed with real diffs and the commit list (bodies included), with a session-transport fallback for free-model-only setups. - Notes: Add to notes distills long selections into 1-3 dense sentences preserving exact identifiers, with verbatim fallback on failure. Fixes along the way: - The global event watcher now starts unconditionally; it was gated behind the desktop-notify env, leaving the server-side event hub dead in packaged apps. - OpenCode re-emits message.updated for old user messages after idle; the watcher no longer mistakes those for new activity. - Session metadata merges from a fresh read right before the PATCH, so writes made during the generation window (suggestion dismissals, review links) are preserved; the assist runtime stops during graceful shutdown. |