You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Default model upgraded to gemini-4-hydrogen: With per-model request timeout raised to 60s (requestTimeoutMs) and overall chain timeout at 180s (timeoutMs).
Customizable fallback chain: Defaults to gemini-4-hydrogen -> gemini-3.8-flash -> gemini-3.5-flash-lite -> gemini-3.7-flash -> gemini-3.1-flash-lite -> gemini-3-flash-preview -> gemini-2.5-flash, configurable via options.fallbackModels or OPENCODE_GOOGLE_FALLBACK_MODELS.
Empty / MAX_TOKENS / blocked fallback: Automatically falls back when a model returns empty output, MAX_TOKENS, or promptFeedback.blockReason, without polluting ctx.storage cache.
🖥️ OpenChamber Web Search UI Integration (ctx.integration)
Registers Google Gemini Grounding (google-grounding) in OpenChamber Settings → Web Search with a configuration form (API Key, optional Base URL, optional Preferred Model) and environment variable detection (OPENCODE_GOOGLE_API_KEY, GOOGLE_API_KEY, GOOGLE_GENERATIVE_AI_API_KEY, GEMINI_API_KEY).
Preserves existing user websearch defaults or disabled preference unless setDefaultWebsearch: true is set or no default is currently selected.
🗄️ OpenCode V2.2.6+ opencode.db & Full Config Resolution
Safe read-only SQLite credential reader: Supports node:sqlite and bun:sqlite with mtime/WAL-aware caching (readCredentialFromSqlite) to read active key / oauth credentials and metadata (baseURL, model) from ~/.local/share/opencode/opencode.db.
Live runtime & project-level resolution: Integrates with live ctx.integration, ctx.provider, and ctx.config, ancestor project opencode.json(c) / .opencode/opencode.json(c), legacy auth.json, GOOGLE_GENERATIVE_AI_API_KEY, and {env:VAR} / {file:path/to/secret} placeholders.
Native thinking preservation & #variant normalization: Omits thinkingConfig by default to keep gemini-4-hydrogen's native server-side HIGH thinking; optionally injects generationConfig.thinkingConfig.thinkingLevel (LOW / MEDIUM / HIGH / XHIGH) for Gemini 3/4 models; normalizeModelId strips both google/ prefixes and #variant suffixes.
Subscriber-aware Singleflight & hardening: Per-caller AbortSignal isolation in Singleflight deduplication; non-default baseURL cache key isolation and expired entry cleanup via storage.remove; immediate FatalError on non-JSON 401/403; error truncation & secret redaction (key=, API keys, Bearer tokens); strict http(s):// card URL sanitization.
Expanded tests: All 23 unit tests passing via npm test.