@mariozechner/pi-ai 0.24.5 → 0.25.0

package/dist/utils/oauth/storage.d.ts

@@ -0,0 +1,81 @@
+/**
+ * OAuth credential storage with configurable backend.
+ *
+ * Default: ~/.pi/agent/oauth.json
+ * Override with setOAuthStorage() for custom storage locations or backends.
+ */
+export interface OAuthCredentials {
+    type: "oauth";
+    refresh: string;
+    access: string;
+    expires: number;
+    enterpriseUrl?: string;
+    projectId?: string;
+    email?: string;
+}
+export interface OAuthStorage {
+    [provider: string]: OAuthCredentials;
+}
+export type OAuthProvider = "anthropic" | "github-copilot" | "google-gemini-cli" | "google-antigravity";
+/**
+ * Storage backend interface.
+ * Implement this to use a custom storage location or backend.
+ */
+export interface OAuthStorageBackend {
+    /** Load all OAuth credentials. Return empty object if none exist. */
+    load(): OAuthStorage;
+    /** Save all OAuth credentials. */
+    save(storage: OAuthStorage): void;
+}
+/**
+ * Configure the OAuth storage backend.
+ *
+ * @example
+ * // Custom file path
+ * setOAuthStorage({
+ *   load: () => JSON.parse(readFileSync('/custom/path/oauth.json', 'utf-8')),
+ *   save: (storage) => writeFileSync('/custom/path/oauth.json', JSON.stringify(storage))
+ * });
+ *
+ * @example
+ * // In-memory storage (for testing)
+ * let memoryStorage = {};
+ * setOAuthStorage({
+ *   load: () => memoryStorage,
+ *   save: (storage) => { memoryStorage = storage; }
+ * });
+ */
+export declare function setOAuthStorage(backend: OAuthStorageBackend): void;
+/**
+ * Reset to default filesystem storage (~/.pi/agent/oauth.json)
+ */
+export declare function resetOAuthStorage(): void;
+/**
+ * Get the default OAuth path (for reference, may not be used if custom backend is set)
+ */
+export declare function getOAuthPath(): string;
+/**
+ * Load all OAuth credentials
+ */
+export declare function loadOAuthStorage(): OAuthStorage;
+/**
+ * Load OAuth credentials for a specific provider
+ */
+export declare function loadOAuthCredentials(provider: string): OAuthCredentials | null;
+/**
+ * Save OAuth credentials for a specific provider
+ */
+export declare function saveOAuthCredentials(provider: string, creds: OAuthCredentials): void;
+/**
+ * Remove OAuth credentials for a specific provider
+ */
+export declare function removeOAuthCredentials(provider: string): void;
+/**
+ * Check if OAuth credentials exist for a provider
+ */
+export declare function hasOAuthCredentials(provider: string): boolean;
+/**
+ * List all providers with OAuth credentials
+ */
+export declare function listOAuthProviders(): string[];
+//# sourceMappingURL=storage.d.ts.map

package/dist/utils/oauth/storage.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"storage.d.ts","sourceRoot":"","sources":["../../../src/utils/oauth/storage.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAMH,MAAM,WAAW,gBAAgB;IAChC,IAAI,EAAE,OAAO,CAAC;IACd,OAAO,EAAE,MAAM,CAAC;IAChB,MAAM,EAAE,MAAM,CAAC;IACf,OAAO,EAAE,MAAM,CAAC;IAChB,aAAa,CAAC,EAAE,MAAM,CAAC;IACvB,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,KAAK,CAAC,EAAE,MAAM,CAAC;CACf;AAED,MAAM,WAAW,YAAY;IAC5B,CAAC,QAAQ,EAAE,MAAM,GAAG,gBAAgB,CAAC;CACrC;AAED,MAAM,MAAM,aAAa,GAAG,WAAW,GAAG,gBAAgB,GAAG,mBAAmB,GAAG,oBAAoB,CAAC;AAExG;;;GAGG;AACH,MAAM,WAAW,mBAAmB;IACnC,qEAAqE;IACrE,IAAI,IAAI,YAAY,CAAC;IACrB,kCAAkC;IAClC,IAAI,CAAC,OAAO,EAAE,YAAY,GAAG,IAAI,CAAC;CAClC;AAsCD;;;;;;;;;;;;;;;;;GAiBG;AACH,wBAAgB,eAAe,CAAC,OAAO,EAAE,mBAAmB,GAAG,IAAI,CAElE;AAED;;GAEG;AACH,wBAAgB,iBAAiB,IAAI,IAAI,CAExC;AAED;;GAEG;AACH,wBAAgB,YAAY,IAAI,MAAM,CAErC;AAMD;;GAEG;AACH,wBAAgB,gBAAgB,IAAI,YAAY,CAE/C;AAED;;GAEG;AACH,wBAAgB,oBAAoB,CAAC,QAAQ,EAAE,MAAM,GAAG,gBAAgB,GAAG,IAAI,CAG9E;AAED;;GAEG;AACH,wBAAgB,oBAAoB,CAAC,QAAQ,EAAE,MAAM,EAAE,KAAK,EAAE,gBAAgB,GAAG,IAAI,CAIpF;AAED;;GAEG;AACH,wBAAgB,sBAAsB,CAAC,QAAQ,EAAE,MAAM,GAAG,IAAI,CAI7D;AAED;;GAEG;AACH,wBAAgB,mBAAmB,CAAC,QAAQ,EAAE,MAAM,GAAG,OAAO,CAE7D;AAED;;GAEG;AACH,wBAAgB,kBAAkB,IAAI,MAAM,EAAE,CAG7C","sourcesContent":["/**\n * OAuth credential storage with configurable backend.\n *\n * Default: ~/.pi/agent/oauth.json\n * Override with setOAuthStorage() for custom storage locations or backends.\n */\n\nimport { chmodSync, existsSync, mkdirSync, readFileSync, writeFileSync } from \"fs\";\nimport { homedir } from \"os\";\nimport { dirname, join } from \"path\";\n\nexport interface OAuthCredentials {\n\ttype: \"oauth\";\n\trefresh: string;\n\taccess: string;\n\texpires: number;\n\tenterpriseUrl?: string;\n\tprojectId?: string;\n\temail?: string;\n}\n\nexport interface OAuthStorage {\n\t[provider: string]: OAuthCredentials;\n}\n\nexport type OAuthProvider = \"anthropic\" | \"github-copilot\" | \"google-gemini-cli\" | \"google-antigravity\";\n\n/**\n * Storage backend interface.\n * Implement this to use a custom storage location or backend.\n */\nexport interface OAuthStorageBackend {\n\t/** Load all OAuth credentials. Return empty object if none exist. */\n\tload(): OAuthStorage;\n\t/** Save all OAuth credentials. */\n\tsave(storage: OAuthStorage): void;\n}\n\n// ============================================================================\n// Default filesystem backend\n// ============================================================================\n\nconst DEFAULT_PATH = join(homedir(), \".pi\", \"agent\", \"oauth.json\");\n\nfunction defaultLoad(): OAuthStorage {\n\tif (!existsSync(DEFAULT_PATH)) {\n\t\treturn {};\n\t}\n\ttry {\n\t\tconst content = readFileSync(DEFAULT_PATH, \"utf-8\");\n\t\treturn JSON.parse(content);\n\t} catch {\n\t\treturn {};\n\t}\n}\n\nfunction defaultSave(storage: OAuthStorage): void {\n\tconst configDir = dirname(DEFAULT_PATH);\n\tif (!existsSync(configDir)) {\n\t\tmkdirSync(configDir, { recursive: true, mode: 0o700 });\n\t}\n\twriteFileSync(DEFAULT_PATH, JSON.stringify(storage, null, 2), \"utf-8\");\n\tchmodSync(DEFAULT_PATH, 0o600);\n}\n\n// ============================================================================\n// Configurable backend\n// ============================================================================\n\nlet currentBackend: OAuthStorageBackend = {\n\tload: defaultLoad,\n\tsave: defaultSave,\n};\n\n/**\n * Configure the OAuth storage backend.\n *\n * @example\n * // Custom file path\n * setOAuthStorage({\n *   load: () => JSON.parse(readFileSync('/custom/path/oauth.json', 'utf-8')),\n *   save: (storage) => writeFileSync('/custom/path/oauth.json', JSON.stringify(storage))\n * });\n *\n * @example\n * // In-memory storage (for testing)\n * let memoryStorage = {};\n * setOAuthStorage({\n *   load: () => memoryStorage,\n *   save: (storage) => { memoryStorage = storage; }\n * });\n */\nexport function setOAuthStorage(backend: OAuthStorageBackend): void {\n\tcurrentBackend = backend;\n}\n\n/**\n * Reset to default filesystem storage (~/.pi/agent/oauth.json)\n */\nexport function resetOAuthStorage(): void {\n\tcurrentBackend = { load: defaultLoad, save: defaultSave };\n}\n\n/**\n * Get the default OAuth path (for reference, may not be used if custom backend is set)\n */\nexport function getOAuthPath(): string {\n\treturn DEFAULT_PATH;\n}\n\n// ============================================================================\n// Public API (uses current backend)\n// ============================================================================\n\n/**\n * Load all OAuth credentials\n */\nexport function loadOAuthStorage(): OAuthStorage {\n\treturn currentBackend.load();\n}\n\n/**\n * Load OAuth credentials for a specific provider\n */\nexport function loadOAuthCredentials(provider: string): OAuthCredentials | null {\n\tconst storage = currentBackend.load();\n\treturn storage[provider] || null;\n}\n\n/**\n * Save OAuth credentials for a specific provider\n */\nexport function saveOAuthCredentials(provider: string, creds: OAuthCredentials): void {\n\tconst storage = currentBackend.load();\n\tstorage[provider] = creds;\n\tcurrentBackend.save(storage);\n}\n\n/**\n * Remove OAuth credentials for a specific provider\n */\nexport function removeOAuthCredentials(provider: string): void {\n\tconst storage = currentBackend.load();\n\tdelete storage[provider];\n\tcurrentBackend.save(storage);\n}\n\n/**\n * Check if OAuth credentials exist for a provider\n */\nexport function hasOAuthCredentials(provider: string): boolean {\n\treturn loadOAuthCredentials(provider) !== null;\n}\n\n/**\n * List all providers with OAuth credentials\n */\nexport function listOAuthProviders(): string[] {\n\tconst storage = currentBackend.load();\n\treturn Object.keys(storage);\n}\n"]}

package/dist/utils/oauth/storage.js

@@ -0,0 +1,119 @@
+/**
+ * OAuth credential storage with configurable backend.
+ *
+ * Default: ~/.pi/agent/oauth.json
+ * Override with setOAuthStorage() for custom storage locations or backends.
+ */
+import { chmodSync, existsSync, mkdirSync, readFileSync, writeFileSync } from "fs";
+import { homedir } from "os";
+import { dirname, join } from "path";
+// ============================================================================
+// Default filesystem backend
+// ============================================================================
+const DEFAULT_PATH = join(homedir(), ".pi", "agent", "oauth.json");
+function defaultLoad() {
+    if (!existsSync(DEFAULT_PATH)) {
+        return {};
+    }
+    try {
+        const content = readFileSync(DEFAULT_PATH, "utf-8");
+        return JSON.parse(content);
+    }
+    catch {
+        return {};
+    }
+}
+function defaultSave(storage) {
+    const configDir = dirname(DEFAULT_PATH);
+    if (!existsSync(configDir)) {
+        mkdirSync(configDir, { recursive: true, mode: 0o700 });
+    }
+    writeFileSync(DEFAULT_PATH, JSON.stringify(storage, null, 2), "utf-8");
+    chmodSync(DEFAULT_PATH, 0o600);
+}
+// ============================================================================
+// Configurable backend
+// ============================================================================
+let currentBackend = {
+    load: defaultLoad,
+    save: defaultSave,
+};
+/**
+ * Configure the OAuth storage backend.
+ *
+ * @example
+ * // Custom file path
+ * setOAuthStorage({
+ *   load: () => JSON.parse(readFileSync('/custom/path/oauth.json', 'utf-8')),
+ *   save: (storage) => writeFileSync('/custom/path/oauth.json', JSON.stringify(storage))
+ * });
+ *
+ * @example
+ * // In-memory storage (for testing)
+ * let memoryStorage = {};
+ * setOAuthStorage({
+ *   load: () => memoryStorage,
+ *   save: (storage) => { memoryStorage = storage; }
+ * });
+ */
+export function setOAuthStorage(backend) {
+    currentBackend = backend;
+}
+/**
+ * Reset to default filesystem storage (~/.pi/agent/oauth.json)
+ */
+export function resetOAuthStorage() {
+    currentBackend = { load: defaultLoad, save: defaultSave };
+}
+/**
+ * Get the default OAuth path (for reference, may not be used if custom backend is set)
+ */
+export function getOAuthPath() {
+    return DEFAULT_PATH;
+}
+// ============================================================================
+// Public API (uses current backend)
+// ============================================================================
+/**
+ * Load all OAuth credentials
+ */
+export function loadOAuthStorage() {
+    return currentBackend.load();
+}
+/**
+ * Load OAuth credentials for a specific provider
+ */
+export function loadOAuthCredentials(provider) {
+    const storage = currentBackend.load();
+    return storage[provider] || null;
+}
+/**
+ * Save OAuth credentials for a specific provider
+ */
+export function saveOAuthCredentials(provider, creds) {
+    const storage = currentBackend.load();
+    storage[provider] = creds;
+    currentBackend.save(storage);
+}
+/**
+ * Remove OAuth credentials for a specific provider
+ */
+export function removeOAuthCredentials(provider) {
+    const storage = currentBackend.load();
+    delete storage[provider];
+    currentBackend.save(storage);
+}
+/**
+ * Check if OAuth credentials exist for a provider
+ */
+export function hasOAuthCredentials(provider) {
+    return loadOAuthCredentials(provider) !== null;
+}
+/**
+ * List all providers with OAuth credentials
+ */
+export function listOAuthProviders() {
+    const storage = currentBackend.load();
+    return Object.keys(storage);
+}
+//# sourceMappingURL=storage.js.map

package/dist/utils/oauth/storage.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"storage.js","sourceRoot":"","sources":["../../../src/utils/oauth/storage.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,EAAE,SAAS,EAAE,UAAU,EAAE,SAAS,EAAE,YAAY,EAAE,aAAa,EAAE,MAAM,IAAI,CAAC;AACnF,OAAO,EAAE,OAAO,EAAE,MAAM,IAAI,CAAC;AAC7B,OAAO,EAAE,OAAO,EAAE,IAAI,EAAE,MAAM,MAAM,CAAC;AA6BrC,+EAA+E;AAC/E,6BAA6B;AAC7B,+EAA+E;AAE/E,MAAM,YAAY,GAAG,IAAI,CAAC,OAAO,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,YAAY,CAAC,CAAC;AAEnE,SAAS,WAAW,GAAiB;IACpC,IAAI,CAAC,UAAU,CAAC,YAAY,CAAC,EAAE,CAAC;QAC/B,OAAO,EAAE,CAAC;IACX,CAAC;IACD,IAAI,CAAC;QACJ,MAAM,OAAO,GAAG,YAAY,CAAC,YAAY,EAAE,OAAO,CAAC,CAAC;QACpD,OAAO,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;IAC5B,CAAC;IAAC,MAAM,CAAC;QACR,OAAO,EAAE,CAAC;IACX,CAAC;AAAA,CACD;AAED,SAAS,WAAW,CAAC,OAAqB,EAAQ;IACjD,MAAM,SAAS,GAAG,OAAO,CAAC,YAAY,CAAC,CAAC;IACxC,IAAI,CAAC,UAAU,CAAC,SAAS,CAAC,EAAE,CAAC;QAC5B,SAAS,CAAC,SAAS,EAAE,EAAE,SAAS,EAAE,IAAI,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC;IACxD,CAAC;IACD,aAAa,CAAC,YAAY,EAAE,IAAI,CAAC,SAAS,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC,CAAC,EAAE,OAAO,CAAC,CAAC;IACvE,SAAS,CAAC,YAAY,EAAE,KAAK,CAAC,CAAC;AAAA,CAC/B;AAED,+EAA+E;AAC/E,uBAAuB;AACvB,+EAA+E;AAE/E,IAAI,cAAc,GAAwB;IACzC,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,WAAW;CACjB,CAAC;AAEF;;;;;;;;;;;;;;;;;GAiBG;AACH,MAAM,UAAU,eAAe,CAAC,OAA4B,EAAQ;IACnE,cAAc,GAAG,OAAO,CAAC;AAAA,CACzB;AAED;;GAEG;AACH,MAAM,UAAU,iBAAiB,GAAS;IACzC,cAAc,GAAG,EAAE,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,WAAW,EAAE,CAAC;AAAA,CAC1D;AAED;;GAEG;AACH,MAAM,UAAU,YAAY,GAAW;IACtC,OAAO,YAAY,CAAC;AAAA,CACpB;AAED,+EAA+E;AAC/E,oCAAoC;AACpC,+EAA+E;AAE/E;;GAEG;AACH,MAAM,UAAU,gBAAgB,GAAiB;IAChD,OAAO,cAAc,CAAC,IAAI,EAAE,CAAC;AAAA,CAC7B;AAED;;GAEG;AACH,MAAM,UAAU,oBAAoB,CAAC,QAAgB,EAA2B;IAC/E,MAAM,OAAO,GAAG,cAAc,CAAC,IAAI,EAAE,CAAC;IACtC,OAAO,OAAO,CAAC,QAAQ,CAAC,IAAI,IAAI,CAAC;AAAA,CACjC;AAED;;GAEG;AACH,MAAM,UAAU,oBAAoB,CAAC,QAAgB,EAAE,KAAuB,EAAQ;IACrF,MAAM,OAAO,GAAG,cAAc,CAAC,IAAI,EAAE,CAAC;IACtC,OAAO,CAAC,QAAQ,CAAC,GAAG,KAAK,CAAC;IAC1B,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;AAAA,CAC7B;AAED;;GAEG;AACH,MAAM,UAAU,sBAAsB,CAAC,QAAgB,EAAQ;IAC9D,MAAM,OAAO,GAAG,cAAc,CAAC,IAAI,EAAE,CAAC;IACtC,OAAO,OAAO,CAAC,QAAQ,CAAC,CAAC;IACzB,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;AAAA,CAC7B;AAED;;GAEG;AACH,MAAM,UAAU,mBAAmB,CAAC,QAAgB,EAAW;IAC9D,OAAO,oBAAoB,CAAC,QAAQ,CAAC,KAAK,IAAI,CAAC;AAAA,CAC/C;AAED;;GAEG;AACH,MAAM,UAAU,kBAAkB,GAAa;IAC9C,MAAM,OAAO,GAAG,cAAc,CAAC,IAAI,EAAE,CAAC;IACtC,OAAO,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;AAAA,CAC5B","sourcesContent":["/**\n * OAuth credential storage with configurable backend.\n *\n * Default: ~/.pi/agent/oauth.json\n * Override with setOAuthStorage() for custom storage locations or backends.\n */\n\nimport { chmodSync, existsSync, mkdirSync, readFileSync, writeFileSync } from \"fs\";\nimport { homedir } from \"os\";\nimport { dirname, join } from \"path\";\n\nexport interface OAuthCredentials {\n\ttype: \"oauth\";\n\trefresh: string;\n\taccess: string;\n\texpires: number;\n\tenterpriseUrl?: string;\n\tprojectId?: string;\n\temail?: string;\n}\n\nexport interface OAuthStorage {\n\t[provider: string]: OAuthCredentials;\n}\n\nexport type OAuthProvider = \"anthropic\" | \"github-copilot\" | \"google-gemini-cli\" | \"google-antigravity\";\n\n/**\n * Storage backend interface.\n * Implement this to use a custom storage location or backend.\n */\nexport interface OAuthStorageBackend {\n\t/** Load all OAuth credentials. Return empty object if none exist. */\n\tload(): OAuthStorage;\n\t/** Save all OAuth credentials. */\n\tsave(storage: OAuthStorage): void;\n}\n\n// ============================================================================\n// Default filesystem backend\n// ============================================================================\n\nconst DEFAULT_PATH = join(homedir(), \".pi\", \"agent\", \"oauth.json\");\n\nfunction defaultLoad(): OAuthStorage {\n\tif (!existsSync(DEFAULT_PATH)) {\n\t\treturn {};\n\t}\n\ttry {\n\t\tconst content = readFileSync(DEFAULT_PATH, \"utf-8\");\n\t\treturn JSON.parse(content);\n\t} catch {\n\t\treturn {};\n\t}\n}\n\nfunction defaultSave(storage: OAuthStorage): void {\n\tconst configDir = dirname(DEFAULT_PATH);\n\tif (!existsSync(configDir)) {\n\t\tmkdirSync(configDir, { recursive: true, mode: 0o700 });\n\t}\n\twriteFileSync(DEFAULT_PATH, JSON.stringify(storage, null, 2), \"utf-8\");\n\tchmodSync(DEFAULT_PATH, 0o600);\n}\n\n// ============================================================================\n// Configurable backend\n// ============================================================================\n\nlet currentBackend: OAuthStorageBackend = {\n\tload: defaultLoad,\n\tsave: defaultSave,\n};\n\n/**\n * Configure the OAuth storage backend.\n *\n * @example\n * // Custom file path\n * setOAuthStorage({\n *   load: () => JSON.parse(readFileSync('/custom/path/oauth.json', 'utf-8')),\n *   save: (storage) => writeFileSync('/custom/path/oauth.json', JSON.stringify(storage))\n * });\n *\n * @example\n * // In-memory storage (for testing)\n * let memoryStorage = {};\n * setOAuthStorage({\n *   load: () => memoryStorage,\n *   save: (storage) => { memoryStorage = storage; }\n * });\n */\nexport function setOAuthStorage(backend: OAuthStorageBackend): void {\n\tcurrentBackend = backend;\n}\n\n/**\n * Reset to default filesystem storage (~/.pi/agent/oauth.json)\n */\nexport function resetOAuthStorage(): void {\n\tcurrentBackend = { load: defaultLoad, save: defaultSave };\n}\n\n/**\n * Get the default OAuth path (for reference, may not be used if custom backend is set)\n */\nexport function getOAuthPath(): string {\n\treturn DEFAULT_PATH;\n}\n\n// ============================================================================\n// Public API (uses current backend)\n// ============================================================================\n\n/**\n * Load all OAuth credentials\n */\nexport function loadOAuthStorage(): OAuthStorage {\n\treturn currentBackend.load();\n}\n\n/**\n * Load OAuth credentials for a specific provider\n */\nexport function loadOAuthCredentials(provider: string): OAuthCredentials | null {\n\tconst storage = currentBackend.load();\n\treturn storage[provider] || null;\n}\n\n/**\n * Save OAuth credentials for a specific provider\n */\nexport function saveOAuthCredentials(provider: string, creds: OAuthCredentials): void {\n\tconst storage = currentBackend.load();\n\tstorage[provider] = creds;\n\tcurrentBackend.save(storage);\n}\n\n/**\n * Remove OAuth credentials for a specific provider\n */\nexport function removeOAuthCredentials(provider: string): void {\n\tconst storage = currentBackend.load();\n\tdelete storage[provider];\n\tcurrentBackend.save(storage);\n}\n\n/**\n * Check if OAuth credentials exist for a provider\n */\nexport function hasOAuthCredentials(provider: string): boolean {\n\treturn loadOAuthCredentials(provider) !== null;\n}\n\n/**\n * List all providers with OAuth credentials\n */\nexport function listOAuthProviders(): string[] {\n\tconst storage = currentBackend.load();\n\treturn Object.keys(storage);\n}\n"]}

package/dist/utils/overflow.d.ts

@@ -16,8 +16,8 @@ import type { AssistantMessage } from "../types.js";
  * - Google Gemini: "input token count exceeds the maximum"
  * - xAI (Grok): "maximum prompt length is X but request contains Y"
  * - Groq: "reduce the length of the messages"
- * - Cerebras: 400/413 status code (no body)
- * - Mistral: 400/413 status code (no body)
+ * - Cerebras: 400/413/429 status code (no body)
+ * - Mistral: 400/413/429 status code (no body)
  * - OpenRouter (all backends): "maximum context length is X tokens"
  * - llama.cpp: "exceeds the available context size"
  * - LM Studio: "greater than the context length"

package/dist/utils/overflow.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"overflow.d.ts","sourceRoot":"","sources":["../../src/utils/overflow.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,aAAa,CAAC;AAqCpD;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA2CG;AACH,wBAAgB,iBAAiB,CAAC,OAAO,EAAE,gBAAgB,EAAE,aAAa,CAAC,EAAE,MAAM,GAAG,OAAO,CAuB5F;AAED;;GAEG;AACH,wBAAgB,mBAAmB,IAAI,MAAM,EAAE,CAE9C","sourcesContent":["import type { AssistantMessage } from \"../types.js\";\n\n/**\n * Regex patterns to detect context overflow errors from different providers.\n *\n * These patterns match error messages returned when the input exceeds\n * the model's context window.\n *\n * Provider-specific patterns (with example error messages):\n *\n * - Anthropic: \"prompt is too long: 213462 tokens > 200000 maximum\"\n * - OpenAI: \"Your input exceeds the context window of this model\"\n * - Google: \"The input token count (1196265) exceeds the maximum number of tokens allowed (1048575)\"\n * - xAI: \"This model's maximum prompt length is 131072 but the request contains 537812 tokens\"\n * - Groq: \"Please reduce the length of the messages or completion\"\n * - OpenRouter: \"This endpoint's maximum context length is X tokens. However, you requested about Y tokens\"\n * - llama.cpp: \"the request exceeds the available context size, try increasing it\"\n * - LM Studio: \"tokens to keep from the initial prompt is greater than the context length\"\n * - Cerebras: Returns \"400 status code (no body)\" - handled separately below\n * - Mistral: Returns \"400 status code (no body)\" - handled separately below\n * - z.ai: Does NOT error, accepts overflow silently - handled via usage.input > contextWindow\n * - Ollama: Silently truncates input - not detectable via error message\n */\nconst OVERFLOW_PATTERNS = [\n\t/prompt is too long/i, // Anthropic\n\t/exceeds the context window/i, // OpenAI (Completions & Responses API)\n\t/input token count.*exceeds the maximum/i, // Google (Gemini)\n\t/maximum prompt length is \\d+/i, // xAI (Grok)\n\t/reduce the length of the messages/i, // Groq\n\t/maximum context length is \\d+ tokens/i, // OpenRouter (all backends)\n\t/exceeds the available context size/i, // llama.cpp server\n\t/greater than the context length/i, // LM Studio\n\t/context length exceeded/i, // Generic fallback\n\t/too many tokens/i, // Generic fallback\n\t/token limit exceeded/i, // Generic fallback\n];\n\n/**\n * Check if an assistant message represents a context overflow error.\n *\n * This handles two cases:\n * 1. Error-based overflow: Most providers return stopReason \"error\" with a\n *    specific error message pattern.\n * 2. Silent overflow: Some providers accept overflow requests and return\n *    successfully. For these, we check if usage.input exceeds the context window.\n *\n * ## Reliability by Provider\n *\n * **Reliable detection (returns error with detectable message):**\n * - Anthropic: \"prompt is too long: X tokens > Y maximum\"\n * - OpenAI (Completions & Responses): \"exceeds the context window\"\n * - Google Gemini: \"input token count exceeds the maximum\"\n * - xAI (Grok): \"maximum prompt length is X but request contains Y\"\n * - Groq: \"reduce the length of the messages\"\n * - Cerebras: 400/413 status code (no body)\n * - Mistral: 400/413 status code (no body)\n * - OpenRouter (all backends): \"maximum context length is X tokens\"\n * - llama.cpp: \"exceeds the available context size\"\n * - LM Studio: \"greater than the context length\"\n *\n * **Unreliable detection:**\n * - z.ai: Sometimes accepts overflow silently (detectable via usage.input > contextWindow),\n *   sometimes returns rate limit errors. Pass contextWindow param to detect silent overflow.\n * - Ollama: Silently truncates input without error. Cannot be detected via this function.\n *   The response will have usage.input < expected, but we don't know the expected value.\n *\n * ## Custom Providers\n *\n * If you've added custom models via settings.json, this function may not detect\n * overflow errors from those providers. To add support:\n *\n * 1. Send a request that exceeds the model's context window\n * 2. Check the errorMessage in the response\n * 3. Create a regex pattern that matches the error\n * 4. The pattern should be added to OVERFLOW_PATTERNS in this file, or\n *    check the errorMessage yourself before calling this function\n *\n * @param message - The assistant message to check\n * @param contextWindow - Optional context window size for detecting silent overflow (z.ai)\n * @returns true if the message indicates a context overflow\n */\nexport function isContextOverflow(message: AssistantMessage, contextWindow?: number): boolean {\n\t// Case 1: Check error message patterns\n\tif (message.stopReason === \"error\" && message.errorMessage) {\n\t\t// Check known patterns\n\t\tif (OVERFLOW_PATTERNS.some((p) => p.test(message.errorMessage!))) {\n\t\t\treturn true;\n\t\t}\n\n\t\t// Cerebras and Mistral return 400/413 with no body - check for status code pattern\n\t\tif (/^4(00|13)\\s*(status code)?\\s*\\(no body\\)/i.test(message.errorMessage)) {\n\t\t\treturn true;\n\t\t}\n\t}\n\n\t// Case 2: Silent overflow (z.ai style) - successful but usage exceeds context\n\tif (contextWindow && message.stopReason === \"stop\") {\n\t\tconst inputTokens = message.usage.input + message.usage.cacheRead;\n\t\tif (inputTokens > contextWindow) {\n\t\t\treturn true;\n\t\t}\n\t}\n\n\treturn false;\n}\n\n/**\n * Get the overflow patterns for testing purposes.\n */\nexport function getOverflowPatterns(): RegExp[] {\n\treturn [...OVERFLOW_PATTERNS];\n}\n"]}
+{"version":3,"file":"overflow.d.ts","sourceRoot":"","sources":["../../src/utils/overflow.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,aAAa,CAAC;AAuCpD;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA2CG;AACH,wBAAgB,iBAAiB,CAAC,OAAO,EAAE,gBAAgB,EAAE,aAAa,CAAC,EAAE,MAAM,GAAG,OAAO,CAwB5F;AAED;;GAEG;AACH,wBAAgB,mBAAmB,IAAI,MAAM,EAAE,CAE9C","sourcesContent":["import type { AssistantMessage } from \"../types.js\";\n\n/**\n * Regex patterns to detect context overflow errors from different providers.\n *\n * These patterns match error messages returned when the input exceeds\n * the model's context window.\n *\n * Provider-specific patterns (with example error messages):\n *\n * - Anthropic: \"prompt is too long: 213462 tokens > 200000 maximum\"\n * - OpenAI: \"Your input exceeds the context window of this model\"\n * - Google: \"The input token count (1196265) exceeds the maximum number of tokens allowed (1048575)\"\n * - xAI: \"This model's maximum prompt length is 131072 but the request contains 537812 tokens\"\n * - Groq: \"Please reduce the length of the messages or completion\"\n * - OpenRouter: \"This endpoint's maximum context length is X tokens. However, you requested about Y tokens\"\n * - llama.cpp: \"the request exceeds the available context size, try increasing it\"\n * - LM Studio: \"tokens to keep from the initial prompt is greater than the context length\"\n * - GitHub Copilot: \"prompt token count of X exceeds the limit of Y\"\n * - Cerebras: Returns \"400 status code (no body)\" - handled separately below\n * - Mistral: Returns \"400 status code (no body)\" - handled separately below\n * - z.ai: Does NOT error, accepts overflow silently - handled via usage.input > contextWindow\n * - Ollama: Silently truncates input - not detectable via error message\n */\nconst OVERFLOW_PATTERNS = [\n\t/prompt is too long/i, // Anthropic\n\t/exceeds the context window/i, // OpenAI (Completions & Responses API)\n\t/input token count.*exceeds the maximum/i, // Google (Gemini)\n\t/maximum prompt length is \\d+/i, // xAI (Grok)\n\t/reduce the length of the messages/i, // Groq\n\t/maximum context length is \\d+ tokens/i, // OpenRouter (all backends)\n\t/exceeds the limit of \\d+/i, // GitHub Copilot\n\t/exceeds the available context size/i, // llama.cpp server\n\t/greater than the context length/i, // LM Studio\n\t/context length exceeded/i, // Generic fallback\n\t/too many tokens/i, // Generic fallback\n\t/token limit exceeded/i, // Generic fallback\n];\n\n/**\n * Check if an assistant message represents a context overflow error.\n *\n * This handles two cases:\n * 1. Error-based overflow: Most providers return stopReason \"error\" with a\n *    specific error message pattern.\n * 2. Silent overflow: Some providers accept overflow requests and return\n *    successfully. For these, we check if usage.input exceeds the context window.\n *\n * ## Reliability by Provider\n *\n * **Reliable detection (returns error with detectable message):**\n * - Anthropic: \"prompt is too long: X tokens > Y maximum\"\n * - OpenAI (Completions & Responses): \"exceeds the context window\"\n * - Google Gemini: \"input token count exceeds the maximum\"\n * - xAI (Grok): \"maximum prompt length is X but request contains Y\"\n * - Groq: \"reduce the length of the messages\"\n * - Cerebras: 400/413/429 status code (no body)\n * - Mistral: 400/413/429 status code (no body)\n * - OpenRouter (all backends): \"maximum context length is X tokens\"\n * - llama.cpp: \"exceeds the available context size\"\n * - LM Studio: \"greater than the context length\"\n *\n * **Unreliable detection:**\n * - z.ai: Sometimes accepts overflow silently (detectable via usage.input > contextWindow),\n *   sometimes returns rate limit errors. Pass contextWindow param to detect silent overflow.\n * - Ollama: Silently truncates input without error. Cannot be detected via this function.\n *   The response will have usage.input < expected, but we don't know the expected value.\n *\n * ## Custom Providers\n *\n * If you've added custom models via settings.json, this function may not detect\n * overflow errors from those providers. To add support:\n *\n * 1. Send a request that exceeds the model's context window\n * 2. Check the errorMessage in the response\n * 3. Create a regex pattern that matches the error\n * 4. The pattern should be added to OVERFLOW_PATTERNS in this file, or\n *    check the errorMessage yourself before calling this function\n *\n * @param message - The assistant message to check\n * @param contextWindow - Optional context window size for detecting silent overflow (z.ai)\n * @returns true if the message indicates a context overflow\n */\nexport function isContextOverflow(message: AssistantMessage, contextWindow?: number): boolean {\n\t// Case 1: Check error message patterns\n\tif (message.stopReason === \"error\" && message.errorMessage) {\n\t\t// Check known patterns\n\t\tif (OVERFLOW_PATTERNS.some((p) => p.test(message.errorMessage!))) {\n\t\t\treturn true;\n\t\t}\n\n\t\t// Cerebras and Mistral return 400/413/429 with no body - check for status code pattern\n\t\t// 429 can indicate token-based rate limiting which correlates with context overflow\n\t\tif (/^4(00|13|29)\\s*(status code)?\\s*\\(no body\\)/i.test(message.errorMessage)) {\n\t\t\treturn true;\n\t\t}\n\t}\n\n\t// Case 2: Silent overflow (z.ai style) - successful but usage exceeds context\n\tif (contextWindow && message.stopReason === \"stop\") {\n\t\tconst inputTokens = message.usage.input + message.usage.cacheRead;\n\t\tif (inputTokens > contextWindow) {\n\t\t\treturn true;\n\t\t}\n\t}\n\n\treturn false;\n}\n\n/**\n * Get the overflow patterns for testing purposes.\n */\nexport function getOverflowPatterns(): RegExp[] {\n\treturn [...OVERFLOW_PATTERNS];\n}\n"]}

package/dist/utils/overflow.js

@@ -14,6 +14,7 @@
  * - OpenRouter: "This endpoint's maximum context length is X tokens. However, you requested about Y tokens"
  * - llama.cpp: "the request exceeds the available context size, try increasing it"
  * - LM Studio: "tokens to keep from the initial prompt is greater than the context length"
+ * - GitHub Copilot: "prompt token count of X exceeds the limit of Y"
  * - Cerebras: Returns "400 status code (no body)" - handled separately below
  * - Mistral: Returns "400 status code (no body)" - handled separately below
  * - z.ai: Does NOT error, accepts overflow silently - handled via usage.input > contextWindow
@@ -26,6 +27,7 @@ const OVERFLOW_PATTERNS = [
     /maximum prompt length is \d+/i, // xAI (Grok)
     /reduce the length of the messages/i, // Groq
     /maximum context length is \d+ tokens/i, // OpenRouter (all backends)
+    /exceeds the limit of \d+/i, // GitHub Copilot
     /exceeds the available context size/i, // llama.cpp server
     /greater than the context length/i, // LM Studio
     /context length exceeded/i, // Generic fallback
@@ -49,8 +51,8 @@ const OVERFLOW_PATTERNS = [
  * - Google Gemini: "input token count exceeds the maximum"
  * - xAI (Grok): "maximum prompt length is X but request contains Y"
  * - Groq: "reduce the length of the messages"
- * - Cerebras: 400/413 status code (no body)
- * - Mistral: 400/413 status code (no body)
+ * - Cerebras: 400/413/429 status code (no body)
+ * - Mistral: 400/413/429 status code (no body)
  * - OpenRouter (all backends): "maximum context length is X tokens"
  * - llama.cpp: "exceeds the available context size"
  * - LM Studio: "greater than the context length"
@@ -83,8 +85,9 @@ export function isContextOverflow(message, contextWindow) {
         if (OVERFLOW_PATTERNS.some((p) => p.test(message.errorMessage))) {
             return true;
         }
-        // Cerebras and Mistral return 400/413 with no body - check for status code pattern
-        if (/^4(00|13)\s*(status code)?\s*\(no body\)/i.test(message.errorMessage)) {
+        // Cerebras and Mistral return 400/413/429 with no body - check for status code pattern
+        // 429 can indicate token-based rate limiting which correlates with context overflow
+        if (/^4(00|13|29)\s*(status code)?\s*\(no body\)/i.test(message.errorMessage)) {
             return true;
         }
     }

package/dist/utils/overflow.js.map

@@ -1 +1 @@
-{"version":3,"file":"overflow.js","sourceRoot":"","sources":["../../src/utils/overflow.ts"],"names":[],"mappings":"AAEA;;;;;;;;;;;;;;;;;;;;GAoBG;AACH,MAAM,iBAAiB,GAAG;IACzB,qBAAqB,EAAE,YAAY;IACnC,6BAA6B,EAAE,uCAAuC;IACtE,yCAAyC,EAAE,kBAAkB;IAC7D,+BAA+B,EAAE,aAAa;IAC9C,oCAAoC,EAAE,OAAO;IAC7C,uCAAuC,EAAE,4BAA4B;IACrE,qCAAqC,EAAE,mBAAmB;IAC1D,kCAAkC,EAAE,YAAY;IAChD,0BAA0B,EAAE,mBAAmB;IAC/C,kBAAkB,EAAE,mBAAmB;IACvC,uBAAuB,EAAE,mBAAmB;CAC5C,CAAC;AAEF;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA2CG;AACH,MAAM,UAAU,iBAAiB,CAAC,OAAyB,EAAE,aAAsB,EAAW;IAC7F,uCAAuC;IACvC,IAAI,OAAO,CAAC,UAAU,KAAK,OAAO,IAAI,OAAO,CAAC,YAAY,EAAE,CAAC;QAC5D,uBAAuB;QACvB,IAAI,iBAAiB,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,YAAa,CAAC,CAAC,EAAE,CAAC;YAClE,OAAO,IAAI,CAAC;QACb,CAAC;QAED,mFAAmF;QACnF,IAAI,2CAA2C,CAAC,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,EAAE,CAAC;YAC5E,OAAO,IAAI,CAAC;QACb,CAAC;IACF,CAAC;IAED,8EAA8E;IAC9E,IAAI,aAAa,IAAI,OAAO,CAAC,UAAU,KAAK,MAAM,EAAE,CAAC;QACpD,MAAM,WAAW,GAAG,OAAO,CAAC,KAAK,CAAC,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,SAAS,CAAC;QAClE,IAAI,WAAW,GAAG,aAAa,EAAE,CAAC;YACjC,OAAO,IAAI,CAAC;QACb,CAAC;IACF,CAAC;IAED,OAAO,KAAK,CAAC;AAAA,CACb;AAED;;GAEG;AACH,MAAM,UAAU,mBAAmB,GAAa;IAC/C,OAAO,CAAC,GAAG,iBAAiB,CAAC,CAAC;AAAA,CAC9B","sourcesContent":["import type { AssistantMessage } from \"../types.js\";\n\n/**\n * Regex patterns to detect context overflow errors from different providers.\n *\n * These patterns match error messages returned when the input exceeds\n * the model's context window.\n *\n * Provider-specific patterns (with example error messages):\n *\n * - Anthropic: \"prompt is too long: 213462 tokens > 200000 maximum\"\n * - OpenAI: \"Your input exceeds the context window of this model\"\n * - Google: \"The input token count (1196265) exceeds the maximum number of tokens allowed (1048575)\"\n * - xAI: \"This model's maximum prompt length is 131072 but the request contains 537812 tokens\"\n * - Groq: \"Please reduce the length of the messages or completion\"\n * - OpenRouter: \"This endpoint's maximum context length is X tokens. However, you requested about Y tokens\"\n * - llama.cpp: \"the request exceeds the available context size, try increasing it\"\n * - LM Studio: \"tokens to keep from the initial prompt is greater than the context length\"\n * - Cerebras: Returns \"400 status code (no body)\" - handled separately below\n * - Mistral: Returns \"400 status code (no body)\" - handled separately below\n * - z.ai: Does NOT error, accepts overflow silently - handled via usage.input > contextWindow\n * - Ollama: Silently truncates input - not detectable via error message\n */\nconst OVERFLOW_PATTERNS = [\n\t/prompt is too long/i, // Anthropic\n\t/exceeds the context window/i, // OpenAI (Completions & Responses API)\n\t/input token count.*exceeds the maximum/i, // Google (Gemini)\n\t/maximum prompt length is \\d+/i, // xAI (Grok)\n\t/reduce the length of the messages/i, // Groq\n\t/maximum context length is \\d+ tokens/i, // OpenRouter (all backends)\n\t/exceeds the available context size/i, // llama.cpp server\n\t/greater than the context length/i, // LM Studio\n\t/context length exceeded/i, // Generic fallback\n\t/too many tokens/i, // Generic fallback\n\t/token limit exceeded/i, // Generic fallback\n];\n\n/**\n * Check if an assistant message represents a context overflow error.\n *\n * This handles two cases:\n * 1. Error-based overflow: Most providers return stopReason \"error\" with a\n *    specific error message pattern.\n * 2. Silent overflow: Some providers accept overflow requests and return\n *    successfully. For these, we check if usage.input exceeds the context window.\n *\n * ## Reliability by Provider\n *\n * **Reliable detection (returns error with detectable message):**\n * - Anthropic: \"prompt is too long: X tokens > Y maximum\"\n * - OpenAI (Completions & Responses): \"exceeds the context window\"\n * - Google Gemini: \"input token count exceeds the maximum\"\n * - xAI (Grok): \"maximum prompt length is X but request contains Y\"\n * - Groq: \"reduce the length of the messages\"\n * - Cerebras: 400/413 status code (no body)\n * - Mistral: 400/413 status code (no body)\n * - OpenRouter (all backends): \"maximum context length is X tokens\"\n * - llama.cpp: \"exceeds the available context size\"\n * - LM Studio: \"greater than the context length\"\n *\n * **Unreliable detection:**\n * - z.ai: Sometimes accepts overflow silently (detectable via usage.input > contextWindow),\n *   sometimes returns rate limit errors. Pass contextWindow param to detect silent overflow.\n * - Ollama: Silently truncates input without error. Cannot be detected via this function.\n *   The response will have usage.input < expected, but we don't know the expected value.\n *\n * ## Custom Providers\n *\n * If you've added custom models via settings.json, this function may not detect\n * overflow errors from those providers. To add support:\n *\n * 1. Send a request that exceeds the model's context window\n * 2. Check the errorMessage in the response\n * 3. Create a regex pattern that matches the error\n * 4. The pattern should be added to OVERFLOW_PATTERNS in this file, or\n *    check the errorMessage yourself before calling this function\n *\n * @param message - The assistant message to check\n * @param contextWindow - Optional context window size for detecting silent overflow (z.ai)\n * @returns true if the message indicates a context overflow\n */\nexport function isContextOverflow(message: AssistantMessage, contextWindow?: number): boolean {\n\t// Case 1: Check error message patterns\n\tif (message.stopReason === \"error\" && message.errorMessage) {\n\t\t// Check known patterns\n\t\tif (OVERFLOW_PATTERNS.some((p) => p.test(message.errorMessage!))) {\n\t\t\treturn true;\n\t\t}\n\n\t\t// Cerebras and Mistral return 400/413 with no body - check for status code pattern\n\t\tif (/^4(00|13)\\s*(status code)?\\s*\\(no body\\)/i.test(message.errorMessage)) {\n\t\t\treturn true;\n\t\t}\n\t}\n\n\t// Case 2: Silent overflow (z.ai style) - successful but usage exceeds context\n\tif (contextWindow && message.stopReason === \"stop\") {\n\t\tconst inputTokens = message.usage.input + message.usage.cacheRead;\n\t\tif (inputTokens > contextWindow) {\n\t\t\treturn true;\n\t\t}\n\t}\n\n\treturn false;\n}\n\n/**\n * Get the overflow patterns for testing purposes.\n */\nexport function getOverflowPatterns(): RegExp[] {\n\treturn [...OVERFLOW_PATTERNS];\n}\n"]}
+{"version":3,"file":"overflow.js","sourceRoot":"","sources":["../../src/utils/overflow.ts"],"names":[],"mappings":"AAEA;;;;;;;;;;;;;;;;;;;;;GAqBG;AACH,MAAM,iBAAiB,GAAG;IACzB,qBAAqB,EAAE,YAAY;IACnC,6BAA6B,EAAE,uCAAuC;IACtE,yCAAyC,EAAE,kBAAkB;IAC7D,+BAA+B,EAAE,aAAa;IAC9C,oCAAoC,EAAE,OAAO;IAC7C,uCAAuC,EAAE,4BAA4B;IACrE,2BAA2B,EAAE,iBAAiB;IAC9C,qCAAqC,EAAE,mBAAmB;IAC1D,kCAAkC,EAAE,YAAY;IAChD,0BAA0B,EAAE,mBAAmB;IAC/C,kBAAkB,EAAE,mBAAmB;IACvC,uBAAuB,EAAE,mBAAmB;CAC5C,CAAC;AAEF;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA2CG;AACH,MAAM,UAAU,iBAAiB,CAAC,OAAyB,EAAE,aAAsB,EAAW;IAC7F,uCAAuC;IACvC,IAAI,OAAO,CAAC,UAAU,KAAK,OAAO,IAAI,OAAO,CAAC,YAAY,EAAE,CAAC;QAC5D,uBAAuB;QACvB,IAAI,iBAAiB,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,CAAC,OAAO,CAAC,YAAa,CAAC,CAAC,EAAE,CAAC;YAClE,OAAO,IAAI,CAAC;QACb,CAAC;QAED,uFAAuF;QACvF,oFAAoF;QACpF,IAAI,8CAA8C,CAAC,IAAI,CAAC,OAAO,CAAC,YAAY,CAAC,EAAE,CAAC;YAC/E,OAAO,IAAI,CAAC;QACb,CAAC;IACF,CAAC;IAED,8EAA8E;IAC9E,IAAI,aAAa,IAAI,OAAO,CAAC,UAAU,KAAK,MAAM,EAAE,CAAC;QACpD,MAAM,WAAW,GAAG,OAAO,CAAC,KAAK,CAAC,KAAK,GAAG,OAAO,CAAC,KAAK,CAAC,SAAS,CAAC;QAClE,IAAI,WAAW,GAAG,aAAa,EAAE,CAAC;YACjC,OAAO,IAAI,CAAC;QACb,CAAC;IACF,CAAC;IAED,OAAO,KAAK,CAAC;AAAA,CACb;AAED;;GAEG;AACH,MAAM,UAAU,mBAAmB,GAAa;IAC/C,OAAO,CAAC,GAAG,iBAAiB,CAAC,CAAC;AAAA,CAC9B","sourcesContent":["import type { AssistantMessage } from \"../types.js\";\n\n/**\n * Regex patterns to detect context overflow errors from different providers.\n *\n * These patterns match error messages returned when the input exceeds\n * the model's context window.\n *\n * Provider-specific patterns (with example error messages):\n *\n * - Anthropic: \"prompt is too long: 213462 tokens > 200000 maximum\"\n * - OpenAI: \"Your input exceeds the context window of this model\"\n * - Google: \"The input token count (1196265) exceeds the maximum number of tokens allowed (1048575)\"\n * - xAI: \"This model's maximum prompt length is 131072 but the request contains 537812 tokens\"\n * - Groq: \"Please reduce the length of the messages or completion\"\n * - OpenRouter: \"This endpoint's maximum context length is X tokens. However, you requested about Y tokens\"\n * - llama.cpp: \"the request exceeds the available context size, try increasing it\"\n * - LM Studio: \"tokens to keep from the initial prompt is greater than the context length\"\n * - GitHub Copilot: \"prompt token count of X exceeds the limit of Y\"\n * - Cerebras: Returns \"400 status code (no body)\" - handled separately below\n * - Mistral: Returns \"400 status code (no body)\" - handled separately below\n * - z.ai: Does NOT error, accepts overflow silently - handled via usage.input > contextWindow\n * - Ollama: Silently truncates input - not detectable via error message\n */\nconst OVERFLOW_PATTERNS = [\n\t/prompt is too long/i, // Anthropic\n\t/exceeds the context window/i, // OpenAI (Completions & Responses API)\n\t/input token count.*exceeds the maximum/i, // Google (Gemini)\n\t/maximum prompt length is \\d+/i, // xAI (Grok)\n\t/reduce the length of the messages/i, // Groq\n\t/maximum context length is \\d+ tokens/i, // OpenRouter (all backends)\n\t/exceeds the limit of \\d+/i, // GitHub Copilot\n\t/exceeds the available context size/i, // llama.cpp server\n\t/greater than the context length/i, // LM Studio\n\t/context length exceeded/i, // Generic fallback\n\t/too many tokens/i, // Generic fallback\n\t/token limit exceeded/i, // Generic fallback\n];\n\n/**\n * Check if an assistant message represents a context overflow error.\n *\n * This handles two cases:\n * 1. Error-based overflow: Most providers return stopReason \"error\" with a\n *    specific error message pattern.\n * 2. Silent overflow: Some providers accept overflow requests and return\n *    successfully. For these, we check if usage.input exceeds the context window.\n *\n * ## Reliability by Provider\n *\n * **Reliable detection (returns error with detectable message):**\n * - Anthropic: \"prompt is too long: X tokens > Y maximum\"\n * - OpenAI (Completions & Responses): \"exceeds the context window\"\n * - Google Gemini: \"input token count exceeds the maximum\"\n * - xAI (Grok): \"maximum prompt length is X but request contains Y\"\n * - Groq: \"reduce the length of the messages\"\n * - Cerebras: 400/413/429 status code (no body)\n * - Mistral: 400/413/429 status code (no body)\n * - OpenRouter (all backends): \"maximum context length is X tokens\"\n * - llama.cpp: \"exceeds the available context size\"\n * - LM Studio: \"greater than the context length\"\n *\n * **Unreliable detection:**\n * - z.ai: Sometimes accepts overflow silently (detectable via usage.input > contextWindow),\n *   sometimes returns rate limit errors. Pass contextWindow param to detect silent overflow.\n * - Ollama: Silently truncates input without error. Cannot be detected via this function.\n *   The response will have usage.input < expected, but we don't know the expected value.\n *\n * ## Custom Providers\n *\n * If you've added custom models via settings.json, this function may not detect\n * overflow errors from those providers. To add support:\n *\n * 1. Send a request that exceeds the model's context window\n * 2. Check the errorMessage in the response\n * 3. Create a regex pattern that matches the error\n * 4. The pattern should be added to OVERFLOW_PATTERNS in this file, or\n *    check the errorMessage yourself before calling this function\n *\n * @param message - The assistant message to check\n * @param contextWindow - Optional context window size for detecting silent overflow (z.ai)\n * @returns true if the message indicates a context overflow\n */\nexport function isContextOverflow(message: AssistantMessage, contextWindow?: number): boolean {\n\t// Case 1: Check error message patterns\n\tif (message.stopReason === \"error\" && message.errorMessage) {\n\t\t// Check known patterns\n\t\tif (OVERFLOW_PATTERNS.some((p) => p.test(message.errorMessage!))) {\n\t\t\treturn true;\n\t\t}\n\n\t\t// Cerebras and Mistral return 400/413/429 with no body - check for status code pattern\n\t\t// 429 can indicate token-based rate limiting which correlates with context overflow\n\t\tif (/^4(00|13|29)\\s*(status code)?\\s*\\(no body\\)/i.test(message.errorMessage)) {\n\t\t\treturn true;\n\t\t}\n\t}\n\n\t// Case 2: Silent overflow (z.ai style) - successful but usage exceeds context\n\tif (contextWindow && message.stopReason === \"stop\") {\n\t\tconst inputTokens = message.usage.input + message.usage.cacheRead;\n\t\tif (inputTokens > contextWindow) {\n\t\t\treturn true;\n\t\t}\n\t}\n\n\treturn false;\n}\n\n/**\n * Get the overflow patterns for testing purposes.\n */\nexport function getOverflowPatterns(): RegExp[] {\n\treturn [...OVERFLOW_PATTERNS];\n}\n"]}

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@mariozechner/pi-ai",
-	"version": "0.24.5",
+	"version": "0.25.0",
 	"description": "Unified LLM API with automatic model discovery and provider configuration",
 	"type": "module",
 	"main": "./dist/index.js",