chore(release): v3.6.6 — Stabilization - #1241
Conversation
…\n\n artifacts - Changed regex quantifier from ? to * in combo.ts, comboAgentMiddleware.ts, and contextHandoff.ts to greedily strip all JSON-escaped newline sequences surrounding <omniModel> tags in SSE streaming chunks - Added \r to the character class for cross-platform robustness - Fixed Playwright strict-mode violation in combo-unification.spec.ts - Bumped OpenAPI version and CHANGELOG to 3.6.6
- fix(gemini): strip VS Code JSON Schema extensions from tool schemas (#1175) Add enumDescriptions, markdownDescription, markdownEnumDescriptions, enumItemLabels and tags to UNSUPPORTED_SCHEMA_CONSTRAINTS so the Gemini sanitizer removes them before forwarding. GitHub Copilot injects these non-standard fields into tool definitions, causing Gemini to reject with 'Unknown name enumDescriptions at functionDeclarations[n].parameters'. - fix(health-check): unwrap proxy config object before passing to getAccessToken (#1187 #1218) resolveProxyForConnection() returns { proxy, level, levelId } but the health check loop was passing the full wrapper to getAccessToken(), which expects the inner config object (.host, .port etc). The proxy dispatcher validated .host on the wrapper (undefined) and threw 'Context proxy host is required', silently marking every connection as unhealthy every sweep. Fix mirrors the pattern already used in chatHelpers.ts: proxyResult?.proxy || null. - fix(ui): debounce models.dev sync interval slider to save only on release (#1202) The slider's onChange fired updateInterval() on every drag tick, sending a PATCH per pixel of movement. Rapid API responses overwrote UI state mid-drag. Introduce draftIntervalHours for smooth visual feedback; the PATCH fires on onMouseUp / onBlur once the user releases the control.
Integrated into release/v3.6.6
…1230) Integrated into release/v3.6.6
…#1228) Integrated into release/v3.6.6
… from PR overlaps
CI Coverage Report
Coverage artifact was not available for this run. |
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 9949fad67b
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| await this.pendingLoad; | ||
| return; | ||
| } | ||
| if (!this.isCacheStale()) return; |
There was a problem hiding this comment.
loadFromDatabase(apiKeyId) is now guarded by a single global freshness check, so once one API key loads skills, subsequent calls for a different API key within 60 seconds return early without querying that tenant's rows. Because handleChatCore now calls skillRegistry.loadFromDatabase(apiKeyInfo.id), this can cause cross-tenant misses where skills are not injected/executed for other API keys until the TTL expires.
Useful? React with 👍 / 👎.
| this._clientHeaders = input.clientHeaders ?? null; | ||
| try { | ||
| const result = await super.execute(input); |
There was a problem hiding this comment.
This stores request-specific headers on a mutable instance field before awaiting super.execute(). Executors are singletons from getExecutor, so concurrent GitHub requests can overwrite _clientHeaders and leak one request's x-initiator into another, causing incorrect upstream X-Initiator values (notably affecting billing semantics for agent vs user).
Useful? React with 👍 / 👎.
|
|
||
| -- Step 6: Recreate triggers using memory_id (INTEGER rowid) instead of id (UUID TEXT) | ||
| CREATE TRIGGER IF NOT EXISTS memory_fts_ai AFTER INSERT ON memories BEGIN | ||
| INSERT INTO memory_fts(rowid, content, key) VALUES (new.memory_id, new.content, new.key); |
There was a problem hiding this comment.
The migration backfills memory_id only once for existing rows, but this trigger indexes future inserts using new.memory_id even though normal inserts do not populate that column. New memories therefore keep memory_id = NULL, so FTS entries get unrelated rowids and JOIN memory_fts ... ON m.memory_id = f.rowid silently drops newly created memories from semantic/hybrid FTS results.
Useful? React with 👍 / 👎.
| const apiKeyId = searchParams.get("apiKeyId") || undefined; | ||
| const type = (searchParams.get("type") as any) || undefined; | ||
| const sessionId = searchParams.get("sessionId") || undefined; | ||
| const limitParams = searchParams.get("limit"); | ||
| const offsetParams = searchParams.get("offset"); | ||
|
|
||
| const memories = await listMemories({ | ||
| const result = await listMemories({ |
There was a problem hiding this comment.
The dashboard now sends q for search and removed client-side filtering, but this route never reads q or forwards any text filter to listMemories. As a result, typing in the memory search box does not narrow results, which is a functional regression in the new paginated flow.
Useful? React with 👍 / 👎.
| rows = db.prepare(ftsQuery).all(...ftsParams) as MemoryRow[]; | ||
| if (rows.length === 0) { |
There was a problem hiding this comment.
In the hybrid branch, FTS query results are assigned to rows, but the merge step unions ftsRows with keyword rows and ftsRows is never populated. This discards all FTS-ranked hits and effectively degrades hybrid retrieval to the fallback chronological/keyword path.
Useful? React with 👍 / 👎.
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request updates OmniRoute to version 3.6.6, introducing a new quota fetcher for Alibaba Coding Plan (Bailian), updating Xiaomi MiMo models, and implementing pagination across memory and skill management interfaces. It also fixes streaming artifacts and adds FTS5-based semantic search for memories. However, several critical issues were identified: a race condition in the GithubExecutor when handling request headers, logic and SQL syntax errors in the hybrid memory retrieval strategy, and a cache isolation bug in the skill registry that could lead to data missing in multi-tenant environments. Additionally, a change in error classification may cause quota exhaustion errors to be misidentified as rate limits, leading to unnecessary retries.
| result.response = new Response(payload, { status, statusText, headers }); | ||
| return result; | ||
| } | ||
| this._clientHeaders = input.clientHeaders ?? null; |
There was a problem hiding this comment.
Storing request-specific headers in a class property (this._clientHeaders) is unsafe if the GithubExecutor instance is shared across concurrent requests (which is typical for executor singletons in this architecture). This creates a race condition where one request's headers can overwrite another's before buildHeaders is called. Consider using AsyncLocalStorage or passing the headers through the call stack if the base class allows it.
| case "hybrid": { | ||
| let ftsRows: MemoryRow[] = []; | ||
| if (config.query && ftsAvailable) { | ||
| const ftsQuery = | ||
| `SELECT m.* FROM ${tableName} m ` + | ||
| `JOIN memory_fts f ON m.memory_id = f.rowid ` + | ||
| `WHERE f.memory_fts MATCH ? AND m.${columns.apiKeyId} = ? ` + | ||
| `AND (m.${columns.expiresAt} IS NULL OR datetime(m.${columns.expiresAt}) > datetime('now'))` + | ||
| (normalizedConfig.scope === "session" && config.sessionId | ||
| ? ` AND m.${columns.sessionId} = ?` | ||
| : "") + | ||
| (normalizedConfig.retentionDays > 0 | ||
| ? ` AND datetime(m.${columns.createdAt}) >= datetime(?)` | ||
| : "") + | ||
| ` ORDER BY f.rank LIMIT 100`; | ||
| const ftsParams: any[] = [config.query, apiKeyId]; | ||
| if (normalizedConfig.scope === "session" && config.sessionId) { | ||
| ftsParams.push(config.sessionId); | ||
| } | ||
| if (normalizedConfig.retentionDays > 0) { | ||
| const cutoff = new Date( | ||
| Date.now() - normalizedConfig.retentionDays * 24 * 60 * 60 * 1000 | ||
| ).toISOString(); | ||
| ftsParams.push(cutoff); | ||
| } | ||
| try { | ||
| rows = db.prepare(ftsQuery).all(...ftsParams) as MemoryRow[]; | ||
| if (rows.length === 0) { | ||
| query += ` ORDER BY ${columns.createdAt} DESC LIMIT 100`; | ||
| rows = db.prepare(query).all(...params) as MemoryRow[]; | ||
| } | ||
| } catch { | ||
| query += ` ORDER BY ${columns.createdAt} DESC LIMIT 100`; | ||
| rows = db.prepare(query).all(...params) as MemoryRow[]; | ||
| } | ||
| } | ||
| // Get chronological results for keyword scoring | ||
| query += ` ORDER BY ${columns.createdAt} DESC LIMIT 100`; | ||
| const keywordRows = db.prepare(query).all(...params) as MemoryRow[]; |
There was a problem hiding this comment.
The hybrid strategy implementation has two significant issues:\n1. Result Assignment Bug: At line 246, the FTS query results are assigned to the shared rows variable instead of ftsRows. Since ftsRows remains empty, the FTS results are ignored during the union at line 263.\n2. SQL Syntax Error Risk: If the FTS query fails or returns no results, the code appends an ORDER BY clause to the query string (lines 249 or 253). However, line 257 appends the same clause again unconditionally. This will result in a malformed SQL statement with duplicate ORDER BY clauses (e.g., SELECT ... ORDER BY ... ORDER BY ...), causing a database error.
| if (this.pendingLoad) { | ||
| await this.pendingLoad; | ||
| return; | ||
| } | ||
| if (!this.isCacheStale()) return; |
There was a problem hiding this comment.
The loadFromDatabase method uses a global lastLoaded timestamp to manage cache TTL, but the loading logic can be filtered by apiKeyId. If the first call to this method is for a specific API key, the registry will only contain skills for that key, and subsequent calls for other keys (or for all skills) will be skipped until the TTL expires. This will cause missing skills for other users in a multi-tenant environment. The cache state should be tracked per apiKeyId, or the method should always perform a full load.
| if (statusCode === 429) { | ||
| return PROVIDER_ERROR_TYPES.RATE_LIMITED; | ||
| } |
There was a problem hiding this comment.
Moving the statusCode === 429 check above the creditsExhausted check changes the behavior for providers that use 429 to signal quota exhaustion (e.g., OpenAI). Previously, these were correctly classified as QUOTA_EXHAUSTED (a terminal error). Now, they will be classified as RATE_LIMITED, which may trigger futile retries by the router. It is generally better to check the response body for specific error codes before falling back to status-code-based classification.
Update package versions for the electron app and open-sse package. Sync llm.txt metadata and feature headings with the 3.6.6 release.
Add guarded outbound fetch helpers with private/local URL blocking, controlled retries, timeout normalization, and route-level status propagation for provider validation and model discovery. Introduce cooldown-aware chat retries with configurable requestRetry and maxRetryIntervalSec settings, model-scoped cooldown responses, and improved rate-limit learning from headers and error bodies so short upstream lockouts can recover automatically. Also align Antigravity and Codex header handling, require API keys for Pollinations, validate web runtime env at startup, restore sanitized Gemini tool names in translated responses, and inject a synthetic Claude text block when upstream SSE completes empty.
Introduce GLM Thinking as a first-class provider preset with shared GLM model metadata, pricing, usage sync, dashboard support, and provider request defaults for higher token budgets and longer timeouts. Use provider-side /messages/count_tokens when a Claude-compatible upstream supports it, while preserving estimated fallback behavior for missing models, missing credentials, and upstream failures. Also add startup seeding for default model aliases and normalize common cross-proxy model dialects so canonical slashful model ids do not get misrouted during resolution.
Add dedicated sync token storage, issuance, revocation, and bundle download routes backed by stable config bundle versioning and ETag support. Expose the v1 websocket handshake route and custom Next server bridge so OpenAI-compatible websocket traffic can be upgraded and proxied through the dashboard and API bridge. Expand compliance auditing with structured metadata, pagination, request context, auth and provider credential events, and SSRF-blocked validation logging.
- CHANGELOG: Add WebSocket bridge, GLM Thinking preset, safe outbound fetch/SSRF guard, cooldown-aware retries, compliance audit v2, model alias seeding, and all Internal Improvements for the 3 new commits - README: Expand v3.6.x highlights table with 10 new features; add SafeOutboundFetch, CooldownAwareRetry, SSRF guard, TPS metric, sync tokens, WebSocket bridge to Resilience/Observability/Deployment tables - ARCHITECTURE: Bump date; add new modules to executive summary, API routes, SSE core services, Auth/Security section; add SSRF/Outbound guard failure mode (section 6); expand module mapping - ENVIRONMENT: Add OMNIROUTE_CRYPT_KEY/OMNIROUTE_API_KEY_BASE64 legacy aliases, OUTBOUND_SSRF_GUARD_ENABLED, CODEX_CLIENT_VERSION, and REQUEST_RETRY/MAX_RETRY_INTERVAL_SEC cooldown retry settings - FEATURES: Add 6 new feature sections — V1 WebSocket Bridge, Sync Tokens & Config Bundle, GLM Thinking Preset, Safe Outbound Fetch & SSRF Guard, Cooldown-Aware Retries, Compliance Audit v2
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 935c36f089
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| const credentials = await getProviderCredentials( | ||
| modelInfo.provider, | ||
| null, | ||
| null, | ||
| modelInfo.model |
There was a problem hiding this comment.
This route now performs real upstream count_tokens calls, but it only checks isValidApiKey and then requests credentials with getProviderCredentials(..., null, null, ...). Passing null for allowedConnections bypasses API-key connection/model restrictions, so any valid key can trigger provider-side counting on accounts it should not access, which can leak model availability and consume restricted upstream quota. Resolve the caller's key metadata and enforce policy constraints before selecting credentials.
Useful? React with 👍 / 👎.
| const result = await listMemories({ | ||
| apiKeyId: args.apiKeyId, | ||
| type: args.type as MemoryType | undefined, | ||
| }); |
There was a problem hiding this comment.
omniroute_memory_clear now deletes only result.data from a single listMemories call, but listMemories is paginated by default (50 rows). For API keys with more than 50 matching memories, this tool reports success after partially clearing only the first page, leaving remaining rows undeleted. The clear path should iterate through pages or request an unbounded result set before deletion.
Useful? React with 👍 / 👎.
Integrated into release/v3.6.6 — IPv6 proxy test fix
#1256) Integrated into release/v3.6.6 — Gemini custom model picker fix
…ests (#1246) Integrated into release/v3.6.6 — OAuth client_id default fallbacks
…n Chat→Responses translator (#1245) Integrated into release/v3.6.6 — max_tokens → max_output_tokens Responses API translation + unit tests
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: da9af4af59
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| const settings = await getSettings().catch(() => ({})); | ||
| return { | ||
| wsAuth: settings.wsAuth === true, |
There was a problem hiding this comment.
If getSettings() throws (for example during transient SQLite failures like SQLITE_BUSY), this code falls back to {} and forces wsAuth to false. In authorizeWebSocketHandshake, that immediately authorizes the handshake without any credential check, so a deployment that has WebSocket auth enabled can become temporarily unauthenticated during DB read failures. This should default to a fail-closed behavior (or return an error) when settings cannot be loaded.
Useful? React with 👍 / 👎.
| this.closed = true; | ||
|
|
||
| clearInterval(this.pingTimer); | ||
| for (const active of this.activeRequests.values()) { | ||
| active.abortController.abort(); | ||
| } | ||
| this.activeRequests.clear(); | ||
|
|
||
| const reasonBuffer = Buffer.from(reason, "utf8"); | ||
| const payload = Buffer.allocUnsafe(2 + reasonBuffer.length); | ||
| payload.writeUInt16BE(code, 0); | ||
| reasonBuffer.copy(payload, 2); | ||
| this.sendFrame(0x8, payload); |
There was a problem hiding this comment.
close() sets this.closed = true before calling sendFrame(0x8, payload), but sendFrame returns early when this.closed is true. That means no close frame is ever sent, so clients can observe abnormal shutdowns (e.g., missing close code/reason) when the server times out or actively closes a session. Reordering this so the close frame is written before flipping the flag avoids protocol-level disconnect issues.
Useful? React with 👍 / 👎.
…eout behavior (#1257) Integrated into release/v3.6.6 — CC-compatible upstream SSE restore + stream timeout fix + README table repair
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: a17b7234fe
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if (isPrivateHost(url.hostname)) { | ||
| throw new OutboundUrlGuardError(PROVIDER_URL_BLOCKED_MESSAGE, { |
There was a problem hiding this comment.
The new public-only guard only checks the literal hostname string, so any domain name that resolves to a private IP (e.g. internal DNS, nip.io-style hosts, or DNS rebinding) bypasses the SSRF protection. Because provider validation/model probes now rely on this guard, a user can still reach loopback/private services through a seemingly public hostname. Add DNS resolution (and post-redirect revalidation) instead of hostname-only checks.
Useful? React with 👍 / 👎.
| this.buffer = Buffer.concat([this.buffer, chunk]); | ||
| const parsed = decodeClientFrames(this.buffer); |
There was a problem hiding this comment.
Incoming WebSocket data is appended with Buffer.concat without any size cap, and frame decoding accepts payload lengths up to Number.MAX_SAFE_INTEGER. A client can send very large or never-finished fragmented frames to force unbounded buffer growth and exhaust memory. This is especially risky on deployments with wsAuth disabled (the new default) or any reachable authenticated WS client. Reject oversized frames and close the socket early.
Useful? React with 👍 / 👎.
| const wsBridge = createOmnirouteWsBridge({ | ||
| baseUrl: `http://127.0.0.1:${dashboardPort}`, | ||
| }); |
There was a problem hiding this comment.
The custom server listens on configurable hostname, but the WS bridge always performs its internal handshake against 127.0.0.1. If HOST is set to a specific non-loopback interface, the handshake fetch targets an address the server is not bound to, so WS upgrades fail even though HTTP serving is healthy. Build the bridge base URL from the actual bound host (or bind to loopback in addition).
Useful? React with 👍 / 👎.
…CLI tools (#1263) Integrated into release/v3.6.6
…riptions, audio-speech) (#1297) Integrated into release/v3.6.6 — adds embeddings, audio-transcriptions, audio-speech, and images-generations support for custom OpenAI-compatible providers, plus Pollinations image registry
There was a problem hiding this comment.
💡 Codex Review
OmniRoute/src/lib/db/migrationRunner.ts
Line 38 in d166362
Fail fast when migration path fallback is not package-rooted
When fileURLToPath(import.meta.url) fails, this fallback points migrations to process.cwd()/src/lib/db/migrations. In global installs or service launches from a different working directory, that folder usually does not exist, so getMigrationFiles() returns an empty list and runMigrations() silently skips all pending migrations. The process can then start against an out-of-date schema and fail later at runtime. This path resolution should be anchored to the installed package location (or throw if migrations cannot be found) instead of relying on CWD.
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Integrated into release/v3.6.6
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: cc867aac7f
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| response: new Response(JSON.stringify(cached), { | ||
| headers: { | ||
| "Content-Type": "application/json", | ||
| "Access-Control-Allow-Origin": getCorsOrigin(), |
There was a problem hiding this comment.
When stream=true, this cache-hit branch returns a JSON payload with Content-Type: application/json instead of an SSE stream. Any OpenAI-compatible client/SDK expecting text/event-stream will fail to parse the response or hang waiting for stream events, so cached responses break the same requests that work on cache misses. The cache path should emit SSE chunks (or bypass cache for streaming requests) to keep response semantics consistent.
Useful? React with 👍 / 👎.
Integrated into release/v3.6.6
|
|
||
| // Support wildcard matching (e.g. 192.168.* or *.local) | ||
| if (patternHost.includes("*")) { | ||
| const regexStr = "^" + patternHost.replace(/\./g, "\\.").replace(/\*/g, ".*") + "$"; |
diegosouzapw
changed the title
Integrated into release/v3.6.6
| try { | ||
| const serialized = JSON.stringify(artifact, null, 2); | ||
| const sizeBytes = Buffer.byteLength(serialized); | ||
| const sha256 = crypto.createHash("sha256").update(serialized).digest("hex"); |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 3d6178870a
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| @@ -486,18 +726,13 @@ export function rotateCallLogs() { | |||
| if (!CALL_LOGS_DIR || !fs.existsSync(CALL_LOGS_DIR)) return; | |||
There was a problem hiding this comment.
rotateCallLogs() now performs SQLite retention (deleteCallLogsBefore and trimCallLogsToMaxRows), but it returns immediately when call_logs/ does not exist. In this release, summary rows are still inserted into call_logs even when no artifact is written (for example noLog traffic or artifact write failures), so this guard skips all DB pruning and allows unbounded table growth until that directory happens to be created.
Useful? React with 👍 / 👎.
| db.prepare("INSERT OR IGNORE INTO _omniroute_migrations (version, name) VALUES (?, ?)").run( | ||
| "026", | ||
| "call_logs_cache_source" |
There was a problem hiding this comment.
This inserts a synthetic 026 migration row even though there is no 026_*.sql file. The migration runner keys applied state by version only, so when a real 026 migration is added later it will be treated as already applied and skipped, causing schema drift on upgrade. Use a non-conflicting marker strategy (or reserve with an actual migration file) instead of pre-populating version 026.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 103a781536
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| return NextResponse.json({ skills }); | ||
| const allSkills = skillRegistry.list(); | ||
| const url = request?.url || "http://localhost/api/skills"; | ||
| const params = parsePaginationParams(new URL(url).searchParams); |
There was a problem hiding this comment.
parsePaginationParams throws on malformed page/limit values (for example GET /api/skills?page=abc), but this handler catches all errors and returns 500, which misclassifies client input errors as server failures. This regression was introduced with the new pagination path and will surface as false backend errors in dashboards/clients sending bad query params.
Useful? React with 👍 / 👎.
| const executions = skillExecutor.listExecutions(); | ||
| return NextResponse.json({ executions }); | ||
| const url = new URL(request.url); | ||
| const params = parsePaginationParams(url.searchParams); |
There was a problem hiding this comment.
The executions list endpoint now parses pagination via parsePaginationParams, which throws on invalid query values (e.g. page=0, limit=foo), but the surrounding catch returns 500 for all exceptions. That makes simple request-validation failures look like internal outages and breaks expected API semantics for bad pagination input.
Useful? React with 👍 / 👎.
- Migrated legacy .mjs scripts to .ts (bin, prepublish, policies) - Resolved pre-commit strict lint (t11 budget) errors in combo.ts - Explicitly typed all TS bindings in pack-artifact policies - Updated package.json commands to run Node via tsx/esm internally - Hardened CI/CD with explicit node version 22.22.2 checks - Completed stage validations for v3.6.6 final release
- Migrate nodeRuntimeSupport to TS interfaces avoiding implicit any - Increase visibility timeouts in skills-marketplace E2E test to 15s to bypass CI flakiness - Complete migration of .mjs scripts to .ts ensuring type safety
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
… 5s to 15s to bypass CI starvation
…g within prepublish hook to fix missing types in CI check
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: e5c16fb196
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| if (mb.progress === "DONE") { | ||
| fullAnswer = chunks.join(""); | ||
| } else { |
There was a problem hiding this comment.
When markdown_block.progress === "DONE", this branch only assigns fullAnswer and does not emit a delta. In the streaming path, buildStreamingResponse() breaks on chunk.done without sending chunk.answer, so requests where Perplexity emits only a final DONE block (or puts new trailing text only in that block) return an empty/truncated assistant message even though upstream produced a valid answer.
Useful? React with 👍 / 👎.
* fix(streaming): diegosouzapw#1211 greedy strip omniModel tags to prevent literal \n\n artifacts - Changed regex quantifier from ? to * in combo.ts, comboAgentMiddleware.ts, and contextHandoff.ts to greedily strip all JSON-escaped newline sequences surrounding <omniModel> tags in SSE streaming chunks - Added \r to the character class for cross-platform robustness - Fixed Playwright strict-mode violation in combo-unification.spec.ts - Bumped OpenAPI version and CHANGELOG to 3.6.6 * fix: 3 bugs found during issue triage (diegosouzapw#1175, diegosouzapw#1187/diegosouzapw#1218, diegosouzapw#1202) - fix(gemini): strip VS Code JSON Schema extensions from tool schemas (diegosouzapw#1175) Add enumDescriptions, markdownDescription, markdownEnumDescriptions, enumItemLabels and tags to UNSUPPORTED_SCHEMA_CONSTRAINTS so the Gemini sanitizer removes them before forwarding. GitHub Copilot injects these non-standard fields into tool definitions, causing Gemini to reject with 'Unknown name enumDescriptions at functionDeclarations[n].parameters'. - fix(health-check): unwrap proxy config object before passing to getAccessToken (diegosouzapw#1187 diegosouzapw#1218) resolveProxyForConnection() returns { proxy, level, levelId } but the health check loop was passing the full wrapper to getAccessToken(), which expects the inner config object (.host, .port etc). The proxy dispatcher validated .host on the wrapper (undefined) and threw 'Context proxy host is required', silently marking every connection as unhealthy every sweep. Fix mirrors the pattern already used in chatHelpers.ts: proxyResult?.proxy || null. - fix(ui): debounce models.dev sync interval slider to save only on release (diegosouzapw#1202) The slider's onChange fired updateInterval() on every drag tick, sending a PATCH per pixel of movement. Rapid API responses overwrote UI state mid-drag. Introduce draftIntervalHours for smooth visual feedback; the PATCH fires on onMouseUp / onBlur once the user releases the control. * fix(providers): update Xiaomi MiMo token-plan endpoints (diegosouzapw#1238) Integrated into release/v3.6.6 * fix(cc-compatible): trim beta flags and preserve cache passthrough (diegosouzapw#1230) Integrated into release/v3.6.6 * feat(memory+skills): full-featured memory & skills systems with tests (diegosouzapw#1228) Integrated into release/v3.6.6 * fix: forward client x-initiator header to GitHub Copilot upstream (diegosouzapw#1227) Integrated into release/v3.6.6 * feat(bailian-quota): add Alibaba Coding Plan quota monitoring (diegosouzapw#1235) * fix: resolve v3.6.6 backlog bugs (diegosouzapw#1206, diegosouzapw#1211, diegosouzapw#1220, diegosouzapw#1231) - fix(core): diegosouzapw#1206 inject startup guard against app/ and src/app/ conflict - fix(health): diegosouzapw#1220 add HEALTHCHECK_STAGGER_MS to prevent token refresh bursting - fix(proxy): diegosouzapw#1231 prioritize HTTP 429 over quota body heuristics - fix(sse): diegosouzapw#1211 strip leading double-newlines in responses API stream * fix(tests): resolve memory migration and skills route pagination bugs from PR overlaps * docs: Update CHANGELOG.md with v3.6.6 features (diegosouzapw#1182, diegosouzapw#1165, diegosouzapw#1177) * chore(release): bump version to 3.6.6 Update package versions for the electron app and open-sse package. Sync llm.txt metadata and feature headings with the 3.6.6 release. * feat(core): harden outbound provider calls and add cooldown retries Add guarded outbound fetch helpers with private/local URL blocking, controlled retries, timeout normalization, and route-level status propagation for provider validation and model discovery. Introduce cooldown-aware chat retries with configurable requestRetry and maxRetryIntervalSec settings, model-scoped cooldown responses, and improved rate-limit learning from headers and error bodies so short upstream lockouts can recover automatically. Also align Antigravity and Codex header handling, require API keys for Pollinations, validate web runtime env at startup, restore sanitized Gemini tool names in translated responses, and inject a synthetic Claude text block when upstream SSE completes empty. * feat(models): add glmt preset and hybrid token counting Introduce GLM Thinking as a first-class provider preset with shared GLM model metadata, pricing, usage sync, dashboard support, and provider request defaults for higher token budgets and longer timeouts. Use provider-side /messages/count_tokens when a Claude-compatible upstream supports it, while preserving estimated fallback behavior for missing models, missing credentials, and upstream failures. Also add startup seeding for default model aliases and normalize common cross-proxy model dialects so canonical slashful model ids do not get misrouted during resolution. * feat(api): add sync tokens and v1 websocket bridge Add dedicated sync token storage, issuance, revocation, and bundle download routes backed by stable config bundle versioning and ETag support. Expose the v1 websocket handshake route and custom Next server bridge so OpenAI-compatible websocket traffic can be upgraded and proxied through the dashboard and API bridge. Expand compliance auditing with structured metadata, pagination, request context, auth and provider credential events, and SSRF-blocked validation logging. * docs: Update all documentation for v3.6.6 - CHANGELOG: Add WebSocket bridge, GLM Thinking preset, safe outbound fetch/SSRF guard, cooldown-aware retries, compliance audit v2, model alias seeding, and all Internal Improvements for the 3 new commits - README: Expand v3.6.x highlights table with 10 new features; add SafeOutboundFetch, CooldownAwareRetry, SSRF guard, TPS metric, sync tokens, WebSocket bridge to Resilience/Observability/Deployment tables - ARCHITECTURE: Bump date; add new modules to executive summary, API routes, SSE core services, Auth/Security section; add SSRF/Outbound guard failure mode (section 6); expand module mapping - ENVIRONMENT: Add OMNIROUTE_CRYPT_KEY/OMNIROUTE_API_KEY_BASE64 legacy aliases, OUTBOUND_SSRF_GUARD_ENABLED, CODEX_CLIENT_VERSION, and REQUEST_RETRY/MAX_RETRY_INTERVAL_SEC cooldown retry settings - FEATURES: Add 6 new feature sections — V1 WebSocket Bridge, Sync Tokens & Config Bundle, GLM Thinking Preset, Safe Outbound Fetch & SSRF Guard, Cooldown-Aware Retries, Compliance Audit v2 * fix: use api64 for proxy test (diegosouzapw#1255) Integrated into release/v3.6.6 — IPv6 proxy test fix * fix(page): update custom models section to include all providers diegosouzapw#1200 (diegosouzapw#1256) Integrated into release/v3.6.6 — Gemini custom model picker fix * fix: provide default client_id fallbacks to prevent broken OAuth requests (diegosouzapw#1246) Integrated into release/v3.6.6 — OAuth client_id default fallbacks * fix: translate max_tokens/max_completion_tokens → max_output_tokens in Chat→Responses translator (diegosouzapw#1245) Integrated into release/v3.6.6 — max_tokens → max_output_tokens Responses API translation + unit tests * feat(oauth): support cursor-agent CLI as Cursor credential source (diegosouzapw#1258) Integrated into release/v3.6.6 — cursor-agent CLI credential source support * fix(cc-compatible): restore upstream SSE and correct stream/combo timeout behavior (diegosouzapw#1257) Integrated into release/v3.6.6 — CC-compatible upstream SSE restore + stream timeout fix + README table repair * fix(cli-tools): resolve API key resolution and model mapping bugs in CLI tools (diegosouzapw#1263) Integrated into release/v3.6.6 * feat(cli-tools): add Qwen Code CLI integration (diegosouzapw#1266) Integrated into release/v3.6.6 * fix(i18n): add missing zh-CN translations and fix logger imports (diegosouzapw#1269) Integrated into release/v3.6.6 * fix(i18n): add Chinese i18n support to dashboard components (diegosouzapw#1274) Integrated into release/v3.6.6 * feat: update Pollinations to require API key, remove free tier flag (diegosouzapw#1177) * feat: friendly error messages for crypto/encryption failures (diegosouzapw#1165) * feat: add TPS (tokens per second) metric column to request logs (diegosouzapw#1182) * feat: merge custom/imported models into filter list for all providers (diegosouzapw#1191) * feat(fallback): Fix provider-profile-driven lockouts (diegosouzapw#1267) This integrates rdself's unify-provider-profile-locks PR manually to handle structural conflicts. * fix(claude): proper Anthropic SDK integration (diegosouzapw#1271) * fix(healthcheck): use correct proxy wrapper format for getAccessToken (diegosouzapw#1272) * chore(release): v3.6.6 — skills registry stability fix + final integration * fix(auth): harden bootstrap auth and memory dashboard behavior Restrict unauthenticated writes to /api/settings/require-login to the initial bootstrap window while keeping read-only checks public. This prevents post-setup config changes without blocking first-run login setup, and the onboarding flow now logs in immediately after setting the password. Restore memory API filtering and pagination behavior by supporting q searches, honoring offset-based requests, and avoiding unrelated fallback results when FTS misses. Update dashboard stats fallback to use the response totals consistently. Package the MCP server with explicit file entries and add regression tests for bootstrap auth and memory route behavior * fix(codex): remove max_output_tokens from body for compatibility * chore(release): v3.6.6 — include PR 1274 fixes in changelog * chore: exclude additional build artifacts and internal directories from npm package distribution * fix: update Gemini OAuth test to match registry defaults + codex UI improvements * fix: restore .mjs refs for scripts/ in test imports after ts migration * fix: restore next.config.mjs ref in dev-origins test * fix: implement db migration safety checks and codex config format * fix: disable mass-migration abort during unit tests based on auto-backup flag * fix: update script regex in auto-update tests to use .mjs * feat: Add Perplexity Web (Session) provider (diegosouzapw#1289) Integrated into release/v3.6.6 * fix(cli): resolve codex routing config parsing, standardize select model button positioning, and clarify oauth documentation * docs(changelog): record recent cli, provider, and test updates Document the latest fixes for Codex routing configuration parsing and Lobehub provider icon fallback behavior. Add the note that the remaining JavaScript test files were migrated to TypeScript ES modules to reflect the completed test stack transition. * chore(release): merge diegosouzapw#1286 minor improvements manually to avoid testing conflict * chore(test): rename perplexity-web.test.mjs to .ts to maintain 100% TS codebase * chore(docs): update CHANGELOG.md for perplexity-web provider * fix(security): resolve CodeQL incomplete URL substring sanitization via URL parsing in test mocks * fix: integrate compressContext() into chatCore.ts request pipeline Proactively compress oversized contexts before sending to upstream providers, preventing context_length_exceeded errors. Compression triggers at 85% of model's context limit using the existing 3-layer compressContext() function. - Import compressContext, estimateTokens, getTokenLimit from contextManager - Add compression check after translation, before executor dispatch - Estimate tokens and compare against 85% threshold of model's context limit - Apply 3-layer compression (trim tools, compress thinking, purify history) - Log compression events with before/after token counts and layers applied - Audit compression events for observability - Add unit tests verifying integration behavior Closes diegosouzapw#1290 * fix(tests): align reasoning expectations with GLM thinking structure * fix: prevent orphaned tool_result messages in purifyHistory() When purifyHistory() drops oldest messages to fit context window, it can split tool_use/tool_result pairs — keeping the tool_result but dropping the tool_use that initiated it. This causes upstream providers to reject the request with format errors. Add fixToolPairs() that runs after each purification pass to remove: - OpenAI format: orphaned role='tool' messages without matching tool_calls ID - Claude format: orphaned tool_result content blocks without matching tool_use ID Closes diegosouzapw#1291 * fix(tests): supply tool_use in mock so it is not dropped * chore: convert remaining test to TypeScript * fix(tests): restore compatibility with compressContext threshold test after tsx migration * docs: finalize v3.6.6 release documentation * fix(core): finalize provider removal, type issues, and codex API key config * fix(dashboard): render Web/Cookie, Search, Audio provider sections and fix TypeScript errors * fix: increase MCP web_search timeout to 60s (diegosouzapw#1278) * fix: route combo testing properly for embedding models (diegosouzapw#1260) * fix: accumulate excluded accounts in combo fallback loop (diegosouzapw#1233) * fix: strip leading whitespace and newlines from first streaming chunk (diegosouzapw#1211) * docs: clarify VPS and Docker settings for OAuth credentials (diegosouzapw#1204) * fix: return real retry-after for pipeline gates (diegosouzapw#1301) Integrated into release/v3.6.6 — returns real Retry-After values from pipeline gates * feat: streaming semantic cache, Cursor auto-version detection, and call-log enhancements (diegosouzapw#1296) Integrated into release/v3.6.6 — streaming semantic cache, Cursor auto-version detection, call-log cache_source tracking * feat(api): support more OpenAI types (image, embeddings, audio-transcriptions, audio-speech) (diegosouzapw#1297) Integrated into release/v3.6.6 — adds embeddings, audio-transcriptions, audio-speech, and images-generations support for custom OpenAI-compatible providers, plus Pollinations image registry * deps: bump hono from 4.12.12 to 4.12.14 (diegosouzapw#1302) Integrated into release/v3.6.6 * deps: bump hono from 4.12.12 to 4.12.14 (diegosouzapw#1306) Integrated into release/v3.6.6 * chore: stabilization fixes for v3.6.6 (diegosouzapw#1298, diegosouzapw#1254, diegosouzapw#59, CI) * fix(providers): match correct endpoint for Xiaomi MiMo, strip routing prefix for custom openai endpoints (diegosouzapw#1303, diegosouzapw#1261) * feat(storage): add database backup cleanup controls * chore(release): v3.6.6 — Final Stabilization Push * Backport call log storage refactor to release/v3.6.6 (diegosouzapw#1307) Integrated into release/v3.6.6 * deps: update dompurify to 3.4.0 to resolve CVE-XYZ (diegosouzapw#60) * test: disable sqlite auto backup in CI to resolve E2E timeout (#24481475058) * chore(docs): sync CHANGELOG for v3.6.6 with missing features and fixes * chore(release): prep v3.6.6 infrastructure and type safety fixes - Migrated legacy .mjs scripts to .ts (bin, prepublish, policies) - Resolved pre-commit strict lint (t11 budget) errors in combo.ts - Explicitly typed all TS bindings in pack-artifact policies - Updated package.json commands to run Node via tsx/esm internally - Hardened CI/CD with explicit node version 22.22.2 checks - Completed stage validations for v3.6.6 final release * chore: fix TS build errors and e2e timeouts in CI - Migrate nodeRuntimeSupport to TS interfaces avoiding implicit any - Increase visibility timeouts in skills-marketplace E2E test to 15s to bypass CI flakiness - Complete migration of .mjs scripts to .ts ensuring type safety * chore(release): sync package version 3.6.6 across workspaces * test(e2e): universally increase UI component visibility timeouts from 5s to 15s to bypass CI starvation * chore(build): inject baseUrl, paths, and types:node into MITM tsconfig within prepublish hook to fix missing types in CI check --------- Co-authored-by: diegosouzapw <diegosouzapw@users.noreply.github.com> Co-authored-by: Jack <5443152+hijak@users.noreply.github.com> Co-authored-by: Randi <55005611+rdself@users.noreply.github.com> Co-authored-by: Paijo <14921983+oyi77@users.noreply.github.com> Co-authored-by: Samuel Cedric <ceds.sam@gmail.com> Co-authored-by: Max Garmash <max@37bytes.com> Co-authored-by: Markus Hartung <mail@hartmark.se> Co-authored-by: Gi99lin <74502520+Gi99lin@users.noreply.github.com> Co-authored-by: Payne <baboialex95@gmail.com> Co-authored-by: Benson K B <bensonkbmca@gmail.com> Co-authored-by: clousky2020 <33016567+clousky2020@users.noreply.github.com> Co-authored-by: Ravi Tharuma <25951435+RaviTharuma@users.noreply.github.com> Co-authored-by: oyi77 <oyi77@users.noreply.github.com> Co-authored-by: Hdsje <vovan877@gmail.com> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com> Co-authored-by: xiaoge1688 <moyekongling@gmail.com>
15 participants
[3.6.6] — 2026-04-15
✨ New Features
/v1/wsWebSocket upgrade route and a custom Next.js server bridge (scripts/v1-ws-bridge.mjs) so OpenAI-compatible WebSocket traffic can be proxied through the gateway. Compliance auditing expanded with structured metadata, pagination, request context, auth/provider credential events, and SSRF-blocked validation logging. New migrations:024_create_sync_tokens.sql. New modules:syncTokens.ts,src/lib/sync/bundle.ts,src/lib/sync/tokens.ts,src/lib/ws/handshake.ts,src/lib/apiBridgeServer.ts,src/lib/compliance/providerAudit.ts.glmt) registered as a first-class provider preset with shared GLM model metadata, pricing, per-connection usage sync, dashboard support, andmaxTokens: 65536 / thinkingBudgetTokens: 24576request defaults with 900s extended timeout. Provider-side/messages/count_tokensendpoint used when a Claude-compatible upstream supports it; gracefully falls back to estimation on missing models, missing credentials, or upstream failures. Startup seeding of default model aliases (src/lib/modelAliasSeed.ts) normalizes common cross-proxy model dialects so canonical slash-based model IDs are not misrouted. New fileopen-sse/config/glmProvider.ts.src/shared/network/safeOutboundFetch.ts,src/shared/network/outboundUrlGuard.ts) blocking private/local URLs with configurable retry, timeout normalisation, and route-level status propagation for provider validation and model discovery. Cooldown-aware chat retries (src/sse/services/cooldownAwareRetry.ts) with configurablerequestRetryandmaxRetryIntervalSecsettings and model-scoped cooldown responses. Improved rate-limit learning from headers and error bodies so short upstream lockouts can recover automatically. Runtime environment validation (src/lib/env/runtimeEnv.ts) checks env at startup. Pollinations now requires an API key. Antigravity and Codex header handling aligned viaopen-sse/config/antigravityUpstream.tsandopen-sse/config/codexClient.ts. Gemini tool names restored in translated responses; synthetic Claude text block injected when upstream SSE completes empty.storage.sqlite) into filesystem artifacts stored withinDATA_DIR/call_logs. This massively reduces WAL bloat and eliminatesSQLITE_FULLcrashes on high-traffic nodes (Backport call log storage refactor to release/v3.6.6 #1307).image,embeddings,audio-transcriptions, andaudio-speechworkflows (feat(api): support more OpenAI types #1297).cursor-agentCLI as a native Cursor credential source alongside the standard configuration (feat(oauth): support cursor-agent CLI as Cursor credential source #1258).🐛 Bug Fixes
fix(providers): match correct endpoint api.xiaomimimo.com for Xiaomi MiMo ([BUG] Incorrect baseurl/endpoint for Xiaomi Mimo #1303)
fix(core): strip provider alias routing prefix from payload for custom endpoints to fix Azure OpenAI 400 errors ([BUG] Azure OpenAI (api-key auth) returns 401 when used via OpenAI-compatible provider #1261)
fix(core): ProxyFetch Undici dispatcher automatically bypasses LAN/local addresses, preventing fetch failures on internal OpenRouter requests (ProxyFetch Undici dispatcher still fails for OpenRouter and OpenAI-compatible LAN providers (v3.6.5) #1254)
fix(core): Gemini thought stream signature detection upgraded to use native part.thought boolean, preventing reasoning text leaks ([BUG] Gemini streaming leaks reasoning into final response #1298)
deps: bump hono from 4.12.12 to 4.12.14 to resolve CVE SSR HTML injection vulnerability (deps: bump hono from 4.12.12 to 4.12.14 #1306, Resolve merge conflict - dependabot branch has grafted history #59)
deps: update dompurify to 3.4.0 in frontend overrides mitigating XSS HTML Injection (CVE-XYZ / Dependabot fix(token-refresh): detect Qwen invalid_request as unrecoverable & switch broken test endpoints to checkExpiry #60)
test: Disable SQLite automatic backups during continuous integration (CI) tests to resolve E2E timeout issues limiting runner scaling (#24481475058)
feat(core): Proactive Context Compression —
chatCorenow proactively compresses oversized message contexts before hitting upstream providers to dramatically reducecontext_length_exceedederrors. Employs binary-search message pruning with structural integrity guarantees tracking explicittool_useboundaries ensuring truncated tool inputs drop paired outputs appropriately (fix: integrate compressContext() into chatCore.ts request pipeline #1292, fix: prevent orphaned tool_result messages in purifyHistory() context compression #1293)fix(cli): Resolve codex routing config parsing by strictly quoting section keys array, enforcing responses wire_api with fallback, and standardizing select-model button positioning mirroring Claude UI
fix(providers): Correct Lobehub provider icons rendering by removing unsupported local references ensuring local SVG/PNG fallback mechanism invokes natively
fix(db): Implement Database migration tracking safety abort safeguards (pre-migration backups via
VACUUM INTOand mass renumbering warnings) to protect existing database structures on startup upgrades (fix: migration safety + 3 bug fixes (MCP timeout, kimi reasoning, GLM empty response) #1281)fix(dashboard): Cleaned up target codex
config.tomlstructure preventing recursive section rendering by enforcing quotes on section dot paths and mapping correct UIOMNIROUTE_API_KEYnames.fix(mcp): Add dedicated explicit timeout constraint overrides for search handlers (fix(mcp): add dedicated timeout for web search #1280)
fix(crypto): Add validation guard to encryption layer to surface clear UI errors when cryptographic environment variables are missing, replacing raw Node.js TypeErrors. Legacy env vars
OMNIROUTE_CRYPT_KEYandOMNIROUTE_API_KEY_BASE64now also accepted as fallbacks (Enhance UI error message for missing OMNIROUTE_CRYPT_KEY crypto failure #1165)fix(providers): Update Pollinations provider definition to require API keys and specify their new limited pollen/hour free tier ([Feature] pollinations changed limits #1177)
Streaming
\n\nArtifact Fix (Leading "\n\n" prefix on assistant responses after v3.5.2 → v3.6.4 upgrade (Chat Completions / combo routing) #1211): Changed<omniModel>tag-stripping regex from?to*quantifier acrosscombo.ts,comboAgentMiddleware.ts, andcontextHandoff.tsto greedily strip all accumulated JSON-escaped newline sequences surrounding the tag. This prevents literal\n\nprefix artifacts from appearing in consumer streaming responsesE2E Combo Test Locator: Fixed Playwright strict-mode violation in
combo-unification.spec.tsby replacing ambiguousgetByRolelocator with a compound filter locator for the "All" strategy tabfix(cc-compatible): Trim beta flags and preserve cache passthrough for third-party HTTP proxy compatibility (fix(cc-compatible): restore conservative beta flags and cache passthrough #1230)
fix(providers): Update Xiaomi MiMo endpoints to the live token-plan, migrating away from dead API URLs (fix(providers): update Xiaomi MiMo token-plan endpoints #1238)
fix: Forward client
x-initiatorheader to GitHub Copilot upstream to accurately distinguish agent vs user turns (fix: forward client x-initiator header to GitHub Copilot upstream #1227)fix: Resolve backlog bugs including streaming edge cases, unhandled rejections, and quota parse failures ([BUG] Gemini CLI. Invakid JSON payload #1206, [Feature] Add configurable stagger delay between token health check sweep iterations #1220, [BUG] "Too Many Requests" (429) is incorrectly treated as "Insufficient Balance / Quota Exhausted", bypassing the Rate Limiting Protection switch and locking the connection #1231, [BUG] UNABLE TO USE GPT AND GEMINI MODELS IN GITHUB COPILOT #1175, [BUG] Network error refreshing Codex token #1187, [BUG] tokenHealthCheck passes full proxy wrapper to getAccessToken instead of unwrapped
.proxyfield #1218, [BUG] Slider bar for "Enable models.dev Sync" is jittery #1202)fix(tests): Resolve memory migration and skills route pagination bugs arising from PR overlaps
fix(i18n): Add missing Chinese i18n support to dashboard components (
DataTable,EmptyState, etc), updateen.json/zh-CN.jsonrouting keys, and natively resolve JSX defaults vianext-intl(fix(i18n): add Chinese i18n support to dashboard components #1274)🔧 Internal Improvements
src/lib/compliance/index.tsexpanded with structured metadata, pagination support, request context enrichment, and newproviderAudit.tsmodule logging auth and provider credential events, SSRF-blocked validation attempts, and provider CRUD operationssrc/lib/sync/bundle.tsexportsbuildConfigBundle()generating a versioned JSON snapshot of settings, provider connections, nodes, model aliases, combos, and API keys (passwords redacted) with ETag support for bandwidth-efficient pollingCODEX_CLIENT_VERSION,CODEX_USER_AGENT_PLATFORM, and pattern-validated env overrides (CODEX_CLIENT_VERSION,CODEX_USER_AGENT) inopen-sse/config/codexClient.tsopen-sse/config/antigravityUpstream.tsconsolidates all Antigravity base URLs and model/fetchAvailableModels discovery path builderssrc/lib/modelAliasSeed.tsseeds 30+ cross-proxy model dialect aliases (e.g.openai/gpt-5→gpt-5,anthropic/claude-opus-4-6→cc/claude-opus-4-6) at startup via idempotentupsertproxy-loadandtestFromFile) to TypeScript ES modules, ensuring a fully synchronized TS stack.models.devauto-sync to combat transient network failures, raised interval floor to 1 hour, and added LKGP debug logging for enhanced observability during routing. (improve: models.dev retry backoff, sync interval floor, LKGP debug logging #1286)