blackriver-gateway 3.8.46

package/open-sse/config/codexIdentity.ts

@@ -0,0 +1,87 @@
+import { createHash, randomUUID } from "node:crypto";
+
+import { normalizeCodexSessionId } from "./codexClient.ts";
+
+const CODEX_INSTALLATION_SALT = "omniroute-codex-installation";
+const UUID_PATTERN = /^[0-9a-f]{8}-[0-9a-f]{4}-[1-5][0-9a-f]{3}-[89ab][0-9a-f]{3}-[0-9a-f]{12}$/i;
+
+export type CodexClientIdentity = {
+  sessionId: string;
+  turnId: string;
+  windowId: string;
+  installationId: string;
+};
+
+function normalizeUuid(value: unknown): string | null {
+  return typeof value === "string" && UUID_PATTERN.test(value.trim()) ? value.trim() : null;
+}
+
+function uuidFromStableValue(value: string): string {
+  const hash = createHash("sha256").update(value).digest("hex");
+  return `${hash.slice(0, 8)}-${hash.slice(8, 12)}-4${hash.slice(13, 16)}-a${hash.slice(17, 20)}-${hash.slice(20, 32)}`;
+}
+
+export function getCodexInstallationId(
+  providerSpecificData?: Record<string, unknown> | null
+): string {
+  const explicit = normalizeUuid(providerSpecificData?.codexInstallationId);
+  if (explicit) return explicit;
+
+  const stableSource =
+    typeof providerSpecificData?.workspaceId === "string" && providerSpecificData.workspaceId.trim()
+      ? providerSpecificData.workspaceId.trim()
+      : typeof providerSpecificData?.accountId === "string" && providerSpecificData.accountId.trim()
+        ? providerSpecificData.accountId.trim()
+        : typeof providerSpecificData?.email === "string" && providerSpecificData.email.trim()
+          ? providerSpecificData.email.trim()
+          : "default";
+
+  return uuidFromStableValue(`${CODEX_INSTALLATION_SALT}:${stableSource}`);
+}
+
+export function createCodexClientIdentity(
+  sessionId: string | null,
+  providerSpecificData?: Record<string, unknown> | null
+): CodexClientIdentity | null {
+  const normalizedSessionId = normalizeCodexSessionId(sessionId);
+  if (!normalizedSessionId) return null;
+  return {
+    sessionId: normalizedSessionId,
+    turnId: randomUUID(),
+    windowId: `${normalizedSessionId}:0`,
+    installationId: getCodexInstallationId(providerSpecificData),
+  };
+}
+
+export function applyCodexClientIdentityHeaders(
+  headers: Record<string, string>,
+  identity?: CodexClientIdentity | null
+): void {
+  if (!identity) return;
+  headers["session_id"] = identity.sessionId;
+  headers["x-client-request-id"] = identity.sessionId;
+  headers["x-codex-window-id"] = identity.windowId;
+  headers["x-codex-turn-metadata"] = JSON.stringify({
+    session_id: identity.sessionId,
+    thread_source: "user",
+    turn_id: identity.turnId,
+    sandbox: "none",
+  });
+}
+
+export function applyCodexClientMetadata(
+  body: Record<string, unknown>,
+  identity?: CodexClientIdentity | null
+): void {
+  if (!identity) return;
+  const existing =
+    body.client_metadata &&
+    typeof body.client_metadata === "object" &&
+    !Array.isArray(body.client_metadata)
+      ? (body.client_metadata as Record<string, unknown>)
+      : {};
+  body.client_metadata = {
+    ...existing,
+    "x-codex-installation-id": identity.installationId,
+  };
+}

package/open-sse/config/codexInstructions.ts

@@ -0,0 +1,122 @@
+// Default instructions for Codex models
+// Source: CLIProxyAPI internal/misc/codex_instructions/
+
+export const CODEX_CHAT_DEFAULT_INSTRUCTIONS = "You are a ChatGPT agent.";
+
+export const CODEX_DEFAULT_INSTRUCTIONS = `You are Codex, based on GPT-5. You are running as a coding agent in the Codex CLI on a user's computer.
+
+## General
+
+- When searching for text or files, prefer using \`rg\` or \`rg --files\` respectively because \`rg\` is much faster than alternatives like \`grep\`. (If the \`rg\` command is not found, then use alternatives.)
+
+## Editing constraints
+
+- Default to ASCII when editing or creating files. Only introduce non-ASCII or other Unicode characters when there is a clear justification and the file already uses them.
+- Add succinct code comments that explain what is going on if code is not self-explanatory. You should not add comments like "Assigns the value to the variable", but a brief comment might be useful ahead of a complex code block that the user would otherwise have to spend time parsing out. Usage of these comments should be rare.
+- Try to use apply_patch for single file edits, but it is fine to explore other options to make the edit if it does not work well. Do not use apply_patch for changes that are auto-generated (i.e. generating package.json or running a lint or format command like gofmt) or when scripting is more efficient (such as search and replacing a string across a codebase).
+- You may be in a dirty git worktree.
+    * NEVER revert existing changes you did not make unless explicitly requested, since these changes were made by the user.
+    * If asked to make a commit or code edits and there are unrelated changes to your work or changes that you didn't make in those files, don't revert those changes.
+    * If the changes are in files you've touched recently, you should read carefully and understand how you can work with the changes rather than reverting them.
+    * If the changes are in unrelated files, just ignore them and don't revert them.
+- Do not amend a commit unless explicitly requested to do so.
+- While you are working, you might notice unexpected changes that you didn't make. If this happens, STOP IMMEDIATELY and ask the user how they would like to proceed.
+- **NEVER** use destructive commands like \`git reset --hard\` or \`git checkout --\` unless specifically requested or approved by the user.
+
+## Plan tool
+
+When using the planning tool:
+- Skip using the planning tool for straightforward tasks (roughly the easiest 25%).
+- Do not make single-step plans.
+- When you made a plan, update it after having performed one of the sub-tasks that you shared on the plan.
+
+## Codex CLI harness, sandboxing, and approvals
+
+The Codex CLI harness supports several different configurations for sandboxing and escalation approvals that the user can choose from.
+
+Filesystem sandboxing defines which files can be read or written. The options for \`sandbox_mode\` are:
+- **read-only**: The sandbox only permits reading files.
+- **workspace-write**: The sandbox permits reading files, and editing files in \`cwd\` and \`writable_roots\`. Editing files in other directories requires approval.
+- **danger-full-access**: No filesystem sandboxing - all commands are permitted.
+
+Network sandboxing defines whether network can be accessed without approval. Options for \`network_access\` are:
+- **restricted**: Requires approval
+- **enabled**: No approval needed
+
+Approvals are your mechanism to get user consent to run shell commands without the sandbox. Possible configuration options for \`approval_policy\` are
+- **untrusted**: The harness will escalate most commands for user approval, apart from a limited allowlist of safe "read" commands.
+- **on-failure**: The harness will allow all commands to run in the sandbox (if enabled), and failures will be escalated to the user for approval to run again without the sandbox.
+- **on-request**: Commands will be run in the sandbox by default, and you can specify in your tool call if you want to escalate a command to run without sandboxing. (Note that this mode is not always available. If it is, you'll see parameters for it in the \`shell\` command description.)
+- **never**: This is a non-interactive mode where you may NEVER ask the user for approval to run commands. Instead, you must always persist and work around constraints to solve the task for the user. You MUST do your utmost best to finish the task and validate your work before yielding. If this mode is paired with \`danger-full-access\`, take advantage of it to deliver the best outcome for the user. Further, in this mode, your default testing philosophy is overridden: Even if you don't see local patterns for testing, you may add tests and scripts to validate your work. Just remove them before yielding.
+
+When you are running with \`approval_policy == on-request\`, and sandboxing enabled, here are scenarios where you'll need to request approval:
+- You need to run a command that writes to a directory that requires it (e.g. running tests that write to /var)
+- You need to run a GUI app (e.g., open/xdg-open/osascript) to open browsers or files.
+- You are running sandboxed and need to run a command that requires network access (e.g. installing packages)
+- If you run a command that is important to solving the user's query, but it fails because of sandboxing, rerun the command with approval. ALWAYS proceed to use the \`sandbox_permissions\` and \`justification\` parameters - do not message the user before requesting approval for the command.
+- You are about to take a potentially destructive action such as an \`rm\` or \`git reset\` that the user did not explicitly ask for
+- (for all of these, you should weigh alternative paths that do not require approval)
+
+When \`sandbox_mode\` is set to read-only, you'll need to request approval for each command that isn't a read.
+
+You will be told what filesystem sandboxing, network sandboxing, and approval mode are active in a developer or user message. If you are not told about this, assume that you are running with workspace-write, network sandboxing enabled, and approval on-failure.
+
+Although they introduce friction to the user because your work is paused until the user responds, you should leverage them when necessary to accomplish important work. If the completing the task requires escalated permissions, Do not let these settings or the sandbox deter you from attempting to accomplish the user's task unless it is set to "never", in which case never ask for approvals.
+
+When requesting approval to execute a command that will require escalated privileges:
+  - Provide the \`sandbox_permissions\` parameter with the value \`"require_escalated"\`
+  - Include a short, 1 sentence explanation for why you need escalated permissions in the justification parameter
+
+## Special user requests
+
+- If the user makes a simple request (such as asking for the time) which you can fulfill by running a terminal command (such as \`date\`), you should do so.
+- If the user asks for a "review", default to a code review mindset: prioritise identifying bugs, risks, behavioural regressions, and missing tests. Findings must be the primary focus of the response - keep summaries or overviews brief and only after enumerating the issues. Present findings first (ordered by severity with file/line references), follow with open questions or assumptions, and offer a change-summary only as a secondary detail. If no findings are discovered, state that explicitly and mention residual risks or testing gaps.
+
+## Frontend tasks
+When doing frontend design tasks, avoid collapsing into "AI slop" or safe, average-looking layouts.
+Aim for interfaces that feel intentional, bold, and a bit surprising.
+- Typography: Use expressive, purposeful fonts and avoid default stacks (Inter, Roboto, Arial, system).
+- Color & Look: Choose a clear visual direction; define CSS variables; avoid purple-on-white defaults. No purple bias or dark mode bias.
+- Motion: Use a few meaningful animations (page-load, staggered reveals) instead of generic micro-motions.
+- Background: Don't rely on flat, single-color backgrounds; use gradients, shapes, or subtle patterns to build atmosphere.
+- Overall: Avoid boilerplate layouts and interchangeable UI patterns. Vary themes, type families, and visual languages across outputs.
+- Ensure the page loads properly on both desktop and mobile
+
+Exception: If working within an existing website or design system, preserve the established patterns, structure, and visual language.
+
+## Presenting your work and final message
+
+You are producing plain text that will later be styled by the CLI. Follow these rules exactly. Formatting should make results easy to scan, but not feel mechanical. Use judgment to decide how much structure adds value.
+
+- Default: be very concise; friendly coding teammate tone.
+- Ask only when needed; suggest ideas; mirror the user's style.
+- For substantial work, summarize clearly; follow final‑answer formatting.
+- Skip heavy formatting for simple confirmations.
+- Don't dump large files you've written; reference paths only.
+- No "save/copy this file" - User is on the same machine.
+- Offer logical next steps (tests, commits, build) briefly; add verify steps if you couldn't do something.
+- For code changes:
+  * Lead with a quick explanation of the change, and then give more details on the context covering where and why a change was made. Do not start this explanation with "summary", just jump right in.
+  * If there are natural next steps the user may want to take, suggest them at the end of your response. Do not make suggestions if there are no natural next steps.
+  * When suggesting multiple options, use numeric lists for the suggestions so the user can quickly respond with a single number.
+- The user does not command execution outputs. When asked to show the output of a command (e.g. \`git show\`), relay the important details in your answer or summarize the key lines so the user understands the result.
+
+### Final answer structure and style guidelines
+
+- Plain text; CLI handles styling. Use structure only when it helps scanability.
+- Headers: optional; short Title Case (1-3 words) wrapped in **…**; no blank line before the first bullet; add only if they truly help.
+- Bullets: use - ; merge related points; keep to one line when possible; 4–6 per list ordered by importance; keep phrasing consistent.
+- Monospace: backticks for commands/paths/env vars/code ids and inline examples; use for literal keyword bullets; never combine with **.
+- Code samples or multi-line snippets should be wrapped in fenced code blocks; include an info string as often as possible.
+- Structure: group related bullets; order sections general → specific → supporting; for subsections, start with a bolded keyword bullet, then items; match complexity to the task.
+- Tone: collaborative, concise, factual; present tense, active voice; self‑contained; no "above/below"; parallel wording.
+- Don'ts: no nested bullets/hierarchies; no ANSI codes; don't cram unrelated keywords; keep keyword lists short—wrap/reformat if long; avoid naming formatting styles in answers.
+- Adaptation: code explanations → precise, structured with code refs; simple tasks → lead with outcome; big changes → logical walkthrough + rationale + next actions; casual one-offs → plain sentences, no headers/bullets.
+- File References: When referencing files in your response follow the below rules:
+  * Use inline code to make file paths clickable.
+  * Each reference should have a stand alone path. Even if it's the same file.
+  * Accepted: absolute, workspace‑relative, a/ or b/ diff prefixes, or bare filename/suffix.
+  * Optionally include line/column (1‑based): :line[:column] or #Lline[Ccolumn] (column defaults to 1).
+  * Do not use URIs like file://, vscode://, or https://.
+  * Do not provide range of lines
+  * Examples: src/app.ts, src/app.ts:42, b/server/index.js#L10, C:\\repo\\project\\main.rs:12:5`;

package/open-sse/config/codexQuotaScopes.ts

@@ -0,0 +1,87 @@
+export type CodexQuotaScope = "codex" | "spark";
+
+export const CODEX_SPARK_MODEL_ID = "gpt-5.3-codex-spark";
+export const CODEX_SPARK_DISPLAY_NAME = "GPT-5.3-Codex-Spark";
+export const CODEX_SPARK_METERED_FEATURE = "gpt_5_3_codex_spark";
+export const CODEX_SPARK_QUOTA_SESSION = `${CODEX_SPARK_METERED_FEATURE}_session`;
+export const CODEX_SPARK_QUOTA_WEEKLY = `${CODEX_SPARK_METERED_FEATURE}_weekly`;
+
+const CODEX_SCOPE_PATTERNS: Array<{ pattern: string; scope: CodexQuotaScope }> = [
+  { pattern: "codex-spark", scope: "spark" },
+  { pattern: "spark", scope: "spark" },
+  { pattern: "bengalfox", scope: "spark" },
+  { pattern: "codex", scope: "codex" },
+  { pattern: "gpt-5", scope: "codex" },
+];
+
+export function getCodexModelScope(model: string | null | undefined): CodexQuotaScope {
+  const lower = String(model || "").toLowerCase();
+  for (const { pattern, scope } of CODEX_SCOPE_PATTERNS) {
+    if (lower.includes(pattern)) return scope;
+  }
+  return "codex";
+}
+
+export function getCodexRateLimitKey(accountId: string, model: string): string {
+  return `${accountId}:${getCodexModelScope(model)}`;
+}
+
+export function isCodexSparkQuotaKey(key: string | null | undefined): boolean {
+  const normalized = String(key || "")
+    .trim()
+    .toLowerCase();
+  if (!normalized) return false;
+  return (
+    normalized === CODEX_SPARK_QUOTA_SESSION ||
+    normalized === CODEX_SPARK_QUOTA_WEEKLY ||
+    normalized === "codex-spark" ||
+    normalized === "codex-spark-weekly" ||
+    normalized.includes("codex-spark") ||
+    normalized.includes("codex_spark") ||
+    normalized.includes(CODEX_SPARK_METERED_FEATURE)
+  );
+}
+
+export function isCodexSparkLimitDescriptor(...values: unknown[]): boolean {
+  return values.some((value) => {
+    if (typeof value !== "string") return false;
+    const normalized = value.trim().toLowerCase();
+    return (
+      normalized.includes("spark") ||
+      normalized.includes("bengalfox") ||
+      normalized.includes(CODEX_SPARK_METERED_FEATURE)
+    );
+  });
+}
+
+export function getCodexQuotaWindowFilterForModel(
+  model: string | null | undefined
+): ((windowName: string) => boolean) | undefined {
+  if (!model) return undefined;
+  const scope = getCodexModelScope(model);
+  return (windowName: string) => {
+    const isSpark = isCodexSparkQuotaKey(windowName);
+    return scope === "spark" ? isSpark : !isSpark;
+  };
+}
+
+export function toCodexScopedQuotaWindowName(
+  baseWindowName: string,
+  model: string | null | undefined
+): string {
+  if (!model || getCodexModelScope(model) !== "spark") return baseWindowName;
+  const normalized = baseWindowName.trim().toLowerCase();
+  if (normalized === "session") return CODEX_SPARK_QUOTA_SESSION;
+  if (normalized === "weekly") return CODEX_SPARK_QUOTA_WEEKLY;
+  return baseWindowName;
+}
+
+export function toCodexBaseQuotaWindowName(windowName: string | null): string | null {
+  if (!windowName) return windowName;
+  const normalized = windowName.trim().toLowerCase();
+  if (normalized === CODEX_SPARK_QUOTA_SESSION || normalized === "codex-spark") return "session";
+  if (normalized === CODEX_SPARK_QUOTA_WEEKLY || normalized === "codex-spark-weekly") {
+    return "weekly";
+  }
+  return windowName;
+}

package/open-sse/config/constants.ts

@@ -0,0 +1,281 @@
+import { getUpstreamTimeoutConfig } from "@/shared/utils/runtimeTimeouts";
+import { loadProviderCredentials } from "./credentialLoader.ts";
+import { generateLegacyProviders } from "./providerRegistry.ts";
+
+const upstreamTimeouts = getUpstreamTimeoutConfig(process.env, (message) => {
+  console.warn(`[open-sse] ${message}`);
+});
+
+// Timeout for receiving the initial upstream response (ms).
+// After headers arrive, active SSE streams are governed by STREAM_IDLE_TIMEOUT_MS
+// and Undici's bodyTimeout instead of this one-shot startup timer.
+export const FETCH_TIMEOUT_MS = upstreamTimeouts.fetchTimeoutMs;
+
+// Idle timeout for SSE streams (ms). Before a stream is accepted, the same
+// budget is used to wait for the first useful event so HTTP 200 zombie streams
+// can fail fast and trigger fallback. After startup, it closes streams that go
+// idle for this duration. Override with STREAM_IDLE_TIMEOUT_MS env var.
+export const STREAM_IDLE_TIMEOUT_MS = upstreamTimeouts.streamIdleTimeoutMs;
+
+// Timeout for the first non-ping SSE event. Inherits REQUEST_TIMEOUT_MS when
+// set, unless STREAM_READINESS_TIMEOUT_MS is specified directly. This must stay
+// conservative for large prompts and slow first-byte reasoning providers.
+export const STREAM_READINESS_TIMEOUT_MS = upstreamTimeouts.streamReadinessTimeoutMs;
+
+// Upper bound for adaptive stream readiness extensions (large histories,
+// tool-heavy requests, high-reasoning Codex targets). Override with
+// STREAM_READINESS_MAX_TIMEOUT_MS when an operator needs longer first-event
+// windows for slow-thinking agent workloads.
+export const STREAM_READINESS_MAX_TIMEOUT_MS = upstreamTimeouts.streamReadinessMaxTimeoutMs;
+
+// Error code used when an upstream Antigravity request stalls before response
+// headers are returned. Keep it shared so executor, core normalization and
+// account fallback detection cannot drift.
+export const ANTIGRAVITY_PRE_RESPONSE_TIMEOUT_CODE = "ANTIGRAVITY_PRE_RESPONSE_TIMEOUT";
+
+// Heartbeat interval for synthetic SSE keepalive emission toward the downstream
+// client (Capy, Claude Code, OpenAI SDK, etc). Keeps strict proxies from
+// dropping the connection during long upstream thinking phases. Set to 0 to
+// disable. Override with SSE_HEARTBEAT_INTERVAL_MS env var.
+export const SSE_HEARTBEAT_INTERVAL_MS = upstreamTimeouts.sseHeartbeatIntervalMs;
+
+// Timeout for reading the full response body after headers arrive (ms).
+// Prevents indefinite hangs when the upstream sends headers but stalls on the body.
+// Defaults to FETCH_TIMEOUT_MS. Override with FETCH_BODY_TIMEOUT_MS env var.
+export const FETCH_BODY_TIMEOUT_MS = upstreamTimeouts.fetchBodyTimeoutMs;
+
+// Provider configurations
+// OAuth credentials read from env vars with hardcoded fallbacks for backward compatibility.
+// Use provider-credentials.json or env vars to override in production.
+export const PROVIDERS = generateLegacyProviders();
+
+// Merge external credentials from data/provider-credentials.json (if present)
+loadProviderCredentials(PROVIDERS);
+
+// Claude system prompt
+export const CLAUDE_SYSTEM_PROMPT = "You are Claude Code, Anthropic's official CLI for Claude.";
+
+// Antigravity default system prompt (required for API to work)
+export const ANTIGRAVITY_DEFAULT_SYSTEM =
+  "You are Antigravity, a powerful agentic AI coding assistant designed by the Google Deepmind team working on Advanced Agentic Coding.\n" +
+  "You are pair programming with a USER to solve their coding task. The task may require creating a new codebase, modifying or debugging an existing codebase, or simply answering a question.\n" +
+  "**Absolute paths only**\n" +
+  "**Proactiveness**";
+
+// OAuth endpoints
+export const OAUTH_ENDPOINTS = {
+  google: {
+    token: "https://oauth2.googleapis.com/token",
+    auth: "https://accounts.google.com/o/oauth2/auth",
+  },
+  openai: {
+    token: "https://auth.openai.com/oauth/token",
+    auth: "https://auth.openai.com/oauth/authorize",
+  },
+  anthropic: {
+    token: "https://api.anthropic.com/v1/oauth/token",
+    auth: "https://api.anthropic.com/v1/oauth/authorize",
+  },
+  qwen: {
+    token: "https://chat.qwen.ai/api/v1/oauth2/token", // From CLIProxyAPI
+    auth: "https://chat.qwen.ai/api/v1/oauth2/device/code", // From CLIProxyAPI
+  },
+  qoder: {
+    token: process.env.QODER_OAUTH_TOKEN_URL || "",
+    auth: process.env.QODER_OAUTH_AUTHORIZE_URL || "",
+  },
+  github: {
+    token: "https://github.com/login/oauth/access_token",
+    auth: "https://github.com/login/oauth/authorize",
+    deviceCode: "https://github.com/login/device/code",
+  },
+};
+
+// Cache TTLs (seconds)
+export const CACHE_TTL = {
+  userInfo: 300, // 5 minutes
+  modelAlias: 3600, // 1 hour
+};
+
+// Default max tokens
+export const DEFAULT_MAX_TOKENS = 64000;
+
+// Minimum max tokens for tool calling (to prevent truncated arguments)
+export const DEFAULT_MIN_TOKENS = 32000;
+
+export const PROVIDER_MAX_TOKENS: Record<string, number> = {
+  groq: 16384, // Groq strict per-model enforcement
+  openai: 16384, // GPT-4/4o standard
+  anthropic: 65536, // Claude models
+  gemini: 65536, // Gemini Studio
+};
+
+export const DEFAULT_PROVIDER_MAX_TOKENS = 32000;
+
+// HTTP status codes
+export const HTTP_STATUS = {
+  BAD_REQUEST: 400,
+  UNAUTHORIZED: 401,
+  PAYMENT_REQUIRED: 402,
+  FORBIDDEN: 403,
+  NOT_FOUND: 404,
+  NOT_ACCEPTABLE: 406,
+  REQUEST_TIMEOUT: 408,
+  RATE_LIMITED: 429,
+  SERVER_ERROR: 500,
+  BAD_GATEWAY: 502,
+  SERVICE_UNAVAILABLE: 503,
+  GATEWAY_TIMEOUT: 504,
+};
+export {
+  BACKOFF_CONFIG,
+  COOLDOWN_MS,
+  DEFAULT_ERROR_MESSAGES,
+  ERROR_RULES,
+  ERROR_TYPES,
+  TRANSIENT_COOLDOWN_MS,
+  calculateBackoffCooldown,
+  findMatchingErrorRule,
+  getDefaultErrorMessage,
+  getErrorInfo,
+  matchErrorRuleByStatus,
+  matchErrorRuleByText,
+} from "./errorConfig.ts";
+
+// Configurable backoff steps for rate limits (Phase 1 — enhanced rate limiting)
+// Used for per-model lockouts with increasing severity
+export const BACKOFF_STEPS_MS = [60_000, 120_000, 300_000, 600_000, 1_200_000];
+// 1min → 2min → 5min → 10min → 20min
+
+// Structured error classification for rate limiting decisions
+export const RateLimitReason = {
+  QUOTA_EXHAUSTED: "quota_exhausted", // Daily/monthly quota depleted
+  RATE_LIMIT_EXCEEDED: "rate_limit_exceeded", // RPM/RPD limits hit
+  MODEL_CAPACITY: "model_capacity", // Model overloaded (529, 503)
+  SERVER_ERROR: "server_error", // 5xx errors
+  AUTH_ERROR: "auth_error", // 401, 403
+  UNKNOWN: "unknown",
+};
+
+// ─── Provider Resilience Profiles ───────────────────────────────────────────
+// Separate behavior for OAuth (low-limit, session-based) vs API Key (high-limit, metered)
+// Circuit-breaker thresholds and reset windows are overridable via
+// OMNIROUTE_CIRCUIT_BREAKER_* env vars so operators can dampen or harden
+// behavior without recompiling.
+function envInt(name: string, fallback: number): number {
+  const raw = process.env[name];
+  if (raw === undefined || raw === null || raw === "") return fallback;
+  const parsed = Number(raw);
+  return Number.isFinite(parsed) && parsed >= 0 ? parsed : fallback;
+}
+
+export const PROVIDER_PROFILES = {
+  oauth: {
+    transientCooldown: 5000, // 5s (session tokens — short recovery)
+    rateLimitCooldown: 60000, // 60s default when no retry-after header
+    maxBackoffLevel: 8, // Higher ceiling (sessions may stay bad longer)
+    circuitBreakerThreshold: envInt("OMNIROUTE_CIRCUIT_BREAKER_OAUTH_THRESHOLD", 8),
+    circuitBreakerReset: envInt("OMNIROUTE_CIRCUIT_BREAKER_OAUTH_RESET_MS", 60000),
+    // Provider-level circuit breaker (entire provider cooldown after repeated failures)
+    providerFailureThreshold: 10, // Scaled for 500+ connections (was 3)
+    providerFailureWindowMs: 900000, // 15min window (was 10min)
+    providerCooldownMs: 300000, // 5min cooldown when threshold reached
+    // Adaptive circuit breaker v2 settings
+    degradationThreshold: 5, // Enter DEGRADED at this many failures
+    maxBackoffMultiplier: 8, // Max 8x resetTimeout escalation
+    backoffEscalationCount: 2, // Escalate after 2 open cycles
+  },
+  apikey: {
+    transientCooldown: 3000, // 3s (API providers recover faster)
+    rateLimitCooldown: 0, // 0 = respect retry-after header from provider
+    maxBackoffLevel: 5, // Lower ceiling (API quotas reset at known intervals)
+    circuitBreakerThreshold: envInt("OMNIROUTE_CIRCUIT_BREAKER_API_KEY_THRESHOLD", 12),
+    circuitBreakerReset: envInt("OMNIROUTE_CIRCUIT_BREAKER_API_KEY_RESET_MS", 30000),
+    // Provider-level circuit breaker (entire provider cooldown after repeated failures)
+    providerFailureThreshold: 15, // Scaled for 500+ connections (was 5)
+    providerFailureWindowMs: 1800000, // 30min window (was 20min)
+    providerCooldownMs: 600000, // 10min cooldown when threshold reached
+    degradationThreshold: 7,
+    maxBackoffMultiplier: 4,
+    backoffEscalationCount: 3,
+  },
+  // Local providers (localhost inference backends like Ollama, LM Studio, oMLX).
+  // Not yet wired into getProviderProfile() — will be used when local provider_nodes
+  // are integrated into the resilience layer. Kept here to avoid a second constants change.
+  local: {
+    transientCooldown: 2000, // 2s (local — very fast recovery)
+    rateLimitCooldown: 5000, // 5s (local — no real rate limits)
+    maxBackoffLevel: 3, // Low ceiling (local either works or doesn't)
+    circuitBreakerThreshold: envInt("OMNIROUTE_CIRCUIT_BREAKER_LOCAL_THRESHOLD", 2),
+    circuitBreakerReset: envInt("OMNIROUTE_CIRCUIT_BREAKER_LOCAL_RESET_MS", 15000),
+    // Provider-level circuit breaker (entire provider cooldown after repeated failures)
+    providerFailureThreshold: 2, // 2 failures trigger provider cooldown
+    providerFailureWindowMs: 300000, // 5min window for counting failures
+    providerCooldownMs: 60000, // 1min cooldown when threshold reached
+  },
+};
+
+// Default rate limit values for API Key providers (auto-enabled safety net)
+// These are intentionally HIGH — they won't restrict normal usage.
+// Real limits are learned from provider response headers.
+export const DEFAULT_API_LIMITS = {
+  requestsPerMinute: 60, // 60 RPM (reduced from 100 — saves Bottleneck queue memory)
+  minTimeBetweenRequests: 350, // 350ms minimum gap (increased from 200)
+  concurrentRequests: 6, // Max 6 parallel per provider (reduced from 10)
+};
+
+// Skip patterns - requests containing these texts will bypass provider
+export const SKIP_PATTERNS = ["Please write a 5-10 word title for the following conversation:"];
+
+// Default maximum number of tools allowed in a request (OpenAI default)
+export const MAX_TOOLS_LIMIT = 128;
+
+// ── Credential Health Check ────────────────────────────────────────
+
+/**
+ * Interval (ms) for the background credential health check scheduler.
+ * Default: 300000 (5 minutes). Minimum: 10000 (10 seconds).
+ */
+export const CREDENTIAL_HEALTH_CHECK_INTERVAL = (() => {
+  const raw = process.env.CREDENTIAL_HEALTH_CHECK_INTERVAL;
+  if (raw) {
+    const parsed = Number(raw);
+    if (Number.isFinite(parsed) && parsed >= 10_000) return parsed;
+  }
+  return 300_000;
+})();
+
+/**
+ * TTL (ms) for cached credential health status.
+ * After this time, the cache entry expires and the next request will
+ * re-check. Default: 300000 (5 minutes).
+ */
+export const CREDENTIAL_HEALTH_CACHE_TTL = (() => {
+  const raw = process.env.CREDENTIAL_HEALTH_CACHE_TTL;
+  if (raw) {
+    const parsed = Number(raw);
+    if (Number.isFinite(parsed) && parsed >= 10_000) return parsed;
+  }
+  return 300_000;
+})();
+
+/**
+ * Stream-recovery tuning (opt-in, see ResilienceSettings.streamRecovery).
+ *
+ * Ported from free-claude-code's always-on recovery (`core/anthropic/stream_recovery.py`).
+ * In OmniRoute the holdback is disabled by default because buffering the opening
+ * window adds up to HOLDBACK_MS of time-to-first-token latency on every stream;
+ * operators opt in via STREAM_RECOVERY_ENABLED / the resilience settings.
+ *
+ * - HOLDBACK_MS: how long the opening SSE window is held so an early truncation
+ *   can be retried transparently before any byte reaches the client.
+ * - BUFFER_MAX_BYTES: hard cap on the held window — commit (flush + passthrough)
+ *   as soon as this many bytes accumulate, regardless of the timer.
+ * - EARLY_RETRY_MAX: max transparent re-opens of the upstream stream while the
+ *   holdback is still uncommitted (free-claude-code uses 5 total attempts = 4 retries).
+ */
+export const STREAM_RECOVERY = {
+  HOLDBACK_MS: 750,
+  BUFFER_MAX_BYTES: 65536,
+  EARLY_RETRY_MAX: 4,
+} as const;