@oh-my-pi/pi-ai 13.3.2 → 13.3.4

package/package.json

@@ -1,7 +1,7 @@
 {
 	"type": "module",
 	"name": "@oh-my-pi/pi-ai",
-	"version": "13.3.2",
+	"version": "13.3.4",
 	"description": "Unified LLM API with automatic model discovery and provider configuration",
 	"homepage": "https://github.com/can1357/oh-my-pi",
 	"author": "Can Boluk",
@@ -38,19 +38,19 @@
 	},
 	"dependencies": {
 		"@anthropic-ai/sdk": "^0.78",
-		"@aws-sdk/client-bedrock-runtime": "^3.995",
+		"@aws-sdk/client-bedrock-runtime": "^3.998",
 		"@bufbuild/protobuf": "^2.11",
 		"@connectrpc/connect": "^2.1",
 		"@connectrpc/connect-node": "^2.1",
-		"@google/genai": "^1.42",
+		"@google/genai": "^1.43",
 		"@mistralai/mistralai": "^1.14",
-		"@oh-my-pi/pi-utils": "13.3.2",
+		"@oh-my-pi/pi-utils": "13.3.4",
 		"@sinclair/typebox": "^0.34",
 		"@smithy/node-http-handler": "^4.4",
 		"ajv": "^8.18",
 		"ajv-formats": "^3.0",
 		"chalk": "^5.6",
-		"openai": "^6.22",
+		"openai": "^6.25",
 		"partial-json": "^0.1",
 		"zod": "^4.3",
 		"zod-to-json-schema": "^3.25"

package/src/models.json

@@ -2664,6 +2664,26 @@
 			"contextWindow": 272000,
 			"maxTokens": 128000
 		},
+		"gpt-5.3-codex": {
+			"id": "gpt-5.3-codex",
+			"name": "GPT-5.3 Codex",
+			"api": "openai-responses",
+			"provider": "opencode",
+			"baseUrl": "https://opencode.ai/zen/v1",
+			"reasoning": true,
+			"input": [
+				"text",
+				"image"
+			],
+			"cost": {
+				"input": 1.75,
+				"output": 14,
+				"cacheRead": 0.175,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 128000
+		},
 		"kimi-k2": {
 			"id": "kimi-k2",
 			"name": "Kimi K2",
@@ -7581,6 +7601,7 @@
 			"contextWindow": 196608,
 			"maxTokens": 128000,
 			"compat": {
+				"supportsStore": false,
 				"supportsDeveloperRole": false,
 				"thinkingFormat": "zai",
 				"reasoningContentField": "reasoning_content"
@@ -7605,6 +7626,7 @@
 			"contextWindow": 204800,
 			"maxTokens": 131072,
 			"compat": {
+				"supportsStore": false,
 				"supportsDeveloperRole": false,
 				"thinkingFormat": "zai",
 				"reasoningContentField": "reasoning_content"
@@ -7653,6 +7675,7 @@
 			"contextWindow": 204800,
 			"maxTokens": 131072,
 			"compat": {
+				"supportsStore": false,
 				"supportsDeveloperRole": false,
 				"thinkingFormat": "zai",
 				"reasoningContentField": "reasoning_content"
@@ -7677,6 +7700,7 @@
 			"contextWindow": 204800,
 			"maxTokens": 131072,
 			"compat": {
+				"supportsStore": false,
 				"supportsDeveloperRole": false,
 				"thinkingFormat": "zai",
 				"reasoningContentField": "reasoning_content"
@@ -7727,6 +7751,7 @@
 			"contextWindow": 196608,
 			"maxTokens": 128000,
 			"compat": {
+				"supportsStore": false,
 				"supportsDeveloperRole": false,
 				"thinkingFormat": "zai",
 				"reasoningContentField": "reasoning_content"
@@ -7751,6 +7776,7 @@
 			"contextWindow": 204800,
 			"maxTokens": 131072,
 			"compat": {
+				"supportsStore": false,
 				"supportsDeveloperRole": false,
 				"thinkingFormat": "zai",
 				"reasoningContentField": "reasoning_content"
@@ -7799,6 +7825,7 @@
 			"contextWindow": 204800,
 			"maxTokens": 131072,
 			"compat": {
+				"supportsStore": false,
 				"supportsDeveloperRole": false,
 				"thinkingFormat": "zai",
 				"reasoningContentField": "reasoning_content"
@@ -7823,6 +7850,7 @@
 			"contextWindow": 204800,
 			"maxTokens": 131072,
 			"compat": {
+				"supportsStore": false,
 				"supportsDeveloperRole": false,
 				"thinkingFormat": "zai",
 				"reasoningContentField": "reasoning_content"
@@ -8105,6 +8133,25 @@
 			"contextWindow": 222222,
 			"maxTokens": 8888
 		},
+		"aion-labs/aion-2.0": {
+			"id": "aion-labs/aion-2.0",
+			"name": "aion-labs/aion-2.0",
+			"api": "openai-completions",
+			"provider": "nanogpt",
+			"baseUrl": "https://nano-gpt.com/api/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 222222,
+			"maxTokens": 8888
+		},
 		"aion-labs/aion-rp-llama-3.1-8b": {
 			"id": "aion-labs/aion-rp-llama-3.1-8b",
 			"name": "aion-labs/aion-rp-llama-3.1-8b",
@@ -12488,6 +12535,25 @@
 			"contextWindow": 222222,
 			"maxTokens": 8888
 		},
+		"liquid/lfm-2-24b-a2b": {
+			"id": "liquid/lfm-2-24b-a2b",
+			"name": "liquid/lfm-2-24b-a2b",
+			"api": "openai-completions",
+			"provider": "nanogpt",
+			"baseUrl": "https://nano-gpt.com/api/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 222222,
+			"maxTokens": 8888
+		},
 		"Llama-3.3-70B-Anthrobomination": {
 			"id": "Llama-3.3-70B-Anthrobomination",
 			"name": "Llama-3.3-70B-Anthrobomination",
@@ -15242,6 +15308,25 @@
 			"contextWindow": 400000,
 			"maxTokens": 128000
 		},
+		"openai/gpt-5.3-codex": {
+			"id": "openai/gpt-5.3-codex",
+			"name": "openai/gpt-5.3-codex",
+			"api": "openai-completions",
+			"provider": "nanogpt",
+			"baseUrl": "https://nano-gpt.com/api/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 8888
+		},
 		"openai/gpt-oss-120b": {
 			"id": "openai/gpt-oss-120b",
 			"name": "GPT OSS 120B",
@@ -16237,6 +16322,158 @@
 			"contextWindow": 222222,
 			"maxTokens": 8888
 		},
+		"qwen3.5-122b-a10b": {
+			"id": "qwen3.5-122b-a10b",
+			"name": "qwen3.5-122b-a10b",
+			"api": "openai-completions",
+			"provider": "nanogpt",
+			"baseUrl": "https://nano-gpt.com/api/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 222222,
+			"maxTokens": 8888
+		},
+		"qwen3.5-122b-a10b:thinking": {
+			"id": "qwen3.5-122b-a10b:thinking",
+			"name": "qwen3.5-122b-a10b:thinking",
+			"api": "openai-completions",
+			"provider": "nanogpt",
+			"baseUrl": "https://nano-gpt.com/api/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 222222,
+			"maxTokens": 8888
+		},
+		"qwen3.5-27b": {
+			"id": "qwen3.5-27b",
+			"name": "qwen3.5-27b",
+			"api": "openai-completions",
+			"provider": "nanogpt",
+			"baseUrl": "https://nano-gpt.com/api/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 222222,
+			"maxTokens": 8888
+		},
+		"qwen3.5-27b:thinking": {
+			"id": "qwen3.5-27b:thinking",
+			"name": "qwen3.5-27b:thinking",
+			"api": "openai-completions",
+			"provider": "nanogpt",
+			"baseUrl": "https://nano-gpt.com/api/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 222222,
+			"maxTokens": 8888
+		},
+		"qwen3.5-35b-a3b": {
+			"id": "qwen3.5-35b-a3b",
+			"name": "qwen3.5-35b-a3b",
+			"api": "openai-completions",
+			"provider": "nanogpt",
+			"baseUrl": "https://nano-gpt.com/api/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 222222,
+			"maxTokens": 8888
+		},
+		"qwen3.5-35b-a3b:thinking": {
+			"id": "qwen3.5-35b-a3b:thinking",
+			"name": "qwen3.5-35b-a3b:thinking",
+			"api": "openai-completions",
+			"provider": "nanogpt",
+			"baseUrl": "https://nano-gpt.com/api/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 222222,
+			"maxTokens": 8888
+		},
+		"qwen3.5-flash": {
+			"id": "qwen3.5-flash",
+			"name": "qwen3.5-flash",
+			"api": "openai-completions",
+			"provider": "nanogpt",
+			"baseUrl": "https://nano-gpt.com/api/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 222222,
+			"maxTokens": 8888
+		},
+		"qwen3.5-flash:thinking": {
+			"id": "qwen3.5-flash:thinking",
+			"name": "qwen3.5-flash:thinking",
+			"api": "openai-completions",
+			"provider": "nanogpt",
+			"baseUrl": "https://nano-gpt.com/api/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 222222,
+			"maxTokens": 8888
+		},
 		"qwq-32b": {
 			"id": "qwq-32b",
 			"name": "qwq-32b",
@@ -19332,13 +19569,13 @@
 				"text"
 			],
 			"cost": {
-				"input": 0.26,
-				"output": 0.38,
+				"input": 0.25,
+				"output": 0.39999999999999997,
 				"cacheRead": 0.13,
 				"cacheWrite": 0
 			},
 			"contextWindow": 163840,
-			"maxTokens": 8888
+			"maxTokens": 163840
 		},
 		"deepseek/deepseek-v3.2-exp": {
 			"id": "deepseek/deepseek-v3.2-exp",
@@ -19599,6 +19836,26 @@
 			"contextWindow": 1048576,
 			"maxTokens": 65536
 		},
+		"google/gemini-3.1-pro-preview-customtools": {
+			"id": "google/gemini-3.1-pro-preview-customtools",
+			"name": "Google: Gemini 3.1 Pro Preview Custom Tools",
+			"api": "openai-completions",
+			"provider": "openrouter",
+			"baseUrl": "https://openrouter.ai/api/v1",
+			"reasoning": true,
+			"input": [
+				"text",
+				"image"
+			],
+			"cost": {
+				"input": 2,
+				"output": 12,
+				"cacheRead": 0.19999999999999998,
+				"cacheWrite": 0.375
+			},
+			"contextWindow": 1048576,
+			"maxTokens": 65536
+		},
 		"google/gemma-3-27b-it": {
 			"id": "google/gemma-3-27b-it",
 			"name": "Gemma-3-27B-IT",
@@ -19696,6 +19953,25 @@
 			"contextWindow": 256000,
 			"maxTokens": 128000
 		},
+		"meituan/longcat-flash-chat": {
+			"id": "meituan/longcat-flash-chat",
+			"name": "Meituan: LongCat Flash Chat",
+			"api": "openai-completions",
+			"provider": "openrouter",
+			"baseUrl": "https://openrouter.ai/api/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0.19999999999999998,
+				"output": 0.7999999999999999,
+				"cacheRead": 0.19999999999999998,
+				"cacheWrite": 0
+			},
+			"contextWindow": 131072,
+			"maxTokens": 32768
+		},
 		"meta-llama/llama-3-8b-instruct": {
 			"id": "meta-llama/llama-3-8b-instruct",
 			"name": "Meta: Llama 3 8B Instruct",
@@ -21371,6 +21647,26 @@
 			"contextWindow": 400000,
 			"maxTokens": 128000
 		},
+		"openai/gpt-5.3-codex": {
+			"id": "openai/gpt-5.3-codex",
+			"name": "OpenAI: GPT-5.3-Codex",
+			"api": "openai-completions",
+			"provider": "openrouter",
+			"baseUrl": "https://openrouter.ai/api/v1",
+			"reasoning": true,
+			"input": [
+				"text",
+				"image"
+			],
+			"cost": {
+				"input": 1.75,
+				"output": 14,
+				"cacheRead": 0.175,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 128000
+		},
 		"openai/gpt-oss-120b": {
 			"id": "openai/gpt-oss-120b",
 			"name": "GPT OSS 120B",
@@ -22471,6 +22767,66 @@
 			"contextWindow": 131072,
 			"maxTokens": 32768
 		},
+		"qwen/qwen3.5-122b-a10b": {
+			"id": "qwen/qwen3.5-122b-a10b",
+			"name": "Qwen: Qwen3.5-122B-A10B",
+			"api": "openai-completions",
+			"provider": "openrouter",
+			"baseUrl": "https://openrouter.ai/api/v1",
+			"reasoning": true,
+			"input": [
+				"text",
+				"image"
+			],
+			"cost": {
+				"input": 0.39999999999999997,
+				"output": 3.1999999999999997,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 262144,
+			"maxTokens": 65536
+		},
+		"qwen/qwen3.5-27b": {
+			"id": "qwen/qwen3.5-27b",
+			"name": "Qwen: Qwen3.5-27B",
+			"api": "openai-completions",
+			"provider": "openrouter",
+			"baseUrl": "https://openrouter.ai/api/v1",
+			"reasoning": true,
+			"input": [
+				"text",
+				"image"
+			],
+			"cost": {
+				"input": 0.3,
+				"output": 2.4,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 262144,
+			"maxTokens": 65536
+		},
+		"qwen/qwen3.5-35b-a3b": {
+			"id": "qwen/qwen3.5-35b-a3b",
+			"name": "Qwen: Qwen3.5-35B-A3B",
+			"api": "openai-completions",
+			"provider": "openrouter",
+			"baseUrl": "https://openrouter.ai/api/v1",
+			"reasoning": true,
+			"input": [
+				"text",
+				"image"
+			],
+			"cost": {
+				"input": 0.25,
+				"output": 2,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 262144,
+			"maxTokens": 65536
+		},
 		"qwen/qwen3.5-397b-a17b": {
 			"id": "qwen/qwen3.5-397b-a17b",
 			"name": "Qwen: Qwen3.5 397B A17B",
@@ -22483,14 +22839,34 @@
 				"image"
 			],
 			"cost": {
-				"input": 0.15,
-				"output": 1,
-				"cacheRead": 0.15,
+				"input": 0.55,
+				"output": 3.5,
+				"cacheRead": 0.55,
 				"cacheWrite": 0
 			},
 			"contextWindow": 262144,
 			"maxTokens": 65536
 		},
+		"qwen/qwen3.5-flash-02-23": {
+			"id": "qwen/qwen3.5-flash-02-23",
+			"name": "Qwen: Qwen3.5-Flash",
+			"api": "openai-completions",
+			"provider": "openrouter",
+			"baseUrl": "https://openrouter.ai/api/v1",
+			"reasoning": true,
+			"input": [
+				"text",
+				"image"
+			],
+			"cost": {
+				"input": 0.09999999999999999,
+				"output": 0.39999999999999997,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 1000000,
+			"maxTokens": 65536
+		},
 		"qwen/qwen3.5-plus-02-15": {
 			"id": "qwen/qwen3.5-plus-02-15",
 			"name": "Qwen: Qwen3.5 Plus 2026-02-15",
@@ -23065,13 +23441,13 @@
 				"text"
 			],
 			"cost": {
-				"input": 0.38,
-				"output": 1.7,
-				"cacheRead": 0.19,
+				"input": 0.3,
+				"output": 1.4,
+				"cacheRead": 0.15,
 				"cacheWrite": 0
 			},
 			"contextWindow": 202752,
-			"maxTokens": 65535
+			"maxTokens": 8888
 		},
 		"z-ai/glm-4.7-flash": {
 			"id": "z-ai/glm-4.7-flash",
@@ -23105,7 +23481,7 @@
 			"cost": {
 				"input": 0.95,
 				"output": 2.5500000000000003,
-				"cacheRead": 0,
+				"cacheRead": 0.19999999999999998,
 				"cacheWrite": 0
 			},
 			"contextWindow": 204800,
@@ -25863,6 +26239,25 @@
 			"contextWindow": 196608,
 			"maxTokens": 8192
 		},
+		"hf:MiniMaxAI/MiniMax-M2.5": {
+			"id": "hf:MiniMaxAI/MiniMax-M2.5",
+			"name": "MiniMaxAI/MiniMax-M2.5",
+			"api": "openai-completions",
+			"provider": "synthetic",
+			"baseUrl": "https://api.synthetic.new/openai/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 191488,
+			"maxTokens": 8192
+		},
 		"hf:moonshotai/Kimi-K2-Instruct-0905": {
 			"id": "hf:moonshotai/Kimi-K2-Instruct-0905",
 			"name": "moonshotai/Kimi-K2-Instruct-0905",
@@ -26464,6 +26859,28 @@
 				"supportsUsageInStreaming": false
 			}
 		},
+		"nvidia-nemotron-3-nano-30b-a3b": {
+			"id": "nvidia-nemotron-3-nano-30b-a3b",
+			"name": "nvidia-nemotron-3-nano-30b-a3b",
+			"api": "openai-completions",
+			"provider": "venice",
+			"baseUrl": "https://api.venice.ai/api/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 222222,
+			"maxTokens": 8888,
+			"compat": {
+				"supportsUsageInStreaming": false
+			}
+		},
 		"olafangensan-glm-4.7-flash-heretic": {
 			"id": "olafangensan-glm-4.7-flash-heretic",
 			"name": "GLM 4.7 Flash Heretic",
@@ -26531,6 +26948,28 @@
 				"supportsUsageInStreaming": false
 			}
 		},
+		"openai-gpt-53-codex": {
+			"id": "openai-gpt-53-codex",
+			"name": "openai-gpt-53-codex",
+			"api": "openai-completions",
+			"provider": "venice",
+			"baseUrl": "https://api.venice.ai/api/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 272000,
+			"maxTokens": 8888,
+			"compat": {
+				"supportsUsageInStreaming": false
+			}
+		},
 		"openai-gpt-oss-120b": {
 			"id": "openai-gpt-oss-120b",
 			"name": "OpenAI GPT OSS 120B",
@@ -26641,6 +27080,28 @@
 				"supportsUsageInStreaming": false
 			}
 		},
+		"qwen3-coder-480b-a35b-instruct-turbo": {
+			"id": "qwen3-coder-480b-a35b-instruct-turbo",
+			"name": "qwen3-coder-480b-a35b-instruct-turbo",
+			"api": "openai-completions",
+			"provider": "venice",
+			"baseUrl": "https://api.venice.ai/api/v1",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 222222,
+			"maxTokens": 8888,
+			"compat": {
+				"supportsUsageInStreaming": false
+			}
+		},
 		"qwen3-next-80b": {
 			"id": "qwen3-next-80b",
 			"name": "Qwen 3 Next 80b",
@@ -28462,6 +28923,25 @@
 			"contextWindow": 272000,
 			"maxTokens": 64000
 		},
+		"gpt-5.3-codex-spark-preview": {
+			"id": "gpt-5.3-codex-spark-preview",
+			"name": "GPT-5.3 Codex Spark",
+			"api": "cursor-agent",
+			"provider": "cursor",
+			"baseUrl": "https://api2.cursor.sh",
+			"reasoning": false,
+			"input": [
+				"text"
+			],
+			"cost": {
+				"input": 0,
+				"output": 0,
+				"cacheRead": 0,
+				"cacheWrite": 0
+			},
+			"contextWindow": 200000,
+			"maxTokens": 64000
+		},
 		"gpt-5.3-codex-xhigh": {
 			"id": "gpt-5.3-codex-xhigh",
 			"name": "GPT-5.3 Codex Extra High",

package/src/provider-models/openai-compat.ts

@@ -1670,6 +1670,7 @@ const MODELS_DEV_PROVIDER_DESCRIPTORS_CODING_PLANS: readonly ModelsDevProviderDe
 	// --- MiniMax Coding Plan ---
 	openAiCompletionsDescriptor("minimax-coding-plan", "minimax-code", "https://api.minimax.io/v1", {
 		compat: {
+			supportsStore: false,
 			supportsDeveloperRole: false,
 			thinkingFormat: "zai",
 			reasoningContentField: "reasoning_content",
@@ -1677,6 +1678,7 @@ const MODELS_DEV_PROVIDER_DESCRIPTORS_CODING_PLANS: readonly ModelsDevProviderDe
 	}),
 	openAiCompletionsDescriptor("minimax-cn-coding-plan", "minimax-code-cn", "https://api.minimaxi.com/v1", {
 		compat: {
+			supportsStore: false,
 			supportsDeveloperRole: false,
 			thinkingFormat: "zai",
 			reasoningContentField: "reasoning_content",

package/src/providers/anthropic.ts

@@ -296,12 +296,18 @@ const PROVIDER_BASE_DELAY_MS = 2000;
  * Includes malformed JSON stream-envelope parse errors seen from some
  * Anthropic-compatible proxy endpoints.
  */
+/** Transient stream corruption errors where the response was truncated mid-JSON. */
+function isTransientStreamParseError(error: unknown): boolean {
+	if (!(error instanceof Error)) return false;
+	return /json parse error|unterminated string|unexpected end of json input/i.test(error.message);
+}
+
 export function isProviderRetryableError(error: unknown): boolean {
 	if (!(error instanceof Error)) return false;
 	const msg = error.message;
 	return (
 		/rate.?limit|too many requests|overloaded|service.?unavailable|1302/i.test(msg) ||
-		/json parse error|unterminated string|unexpected end of json input/i.test(msg)
+		isTransientStreamParseError(error)
 	);
 }
 
@@ -371,9 +377,9 @@ export const streamAnthropic: StreamFunction<"anthropic-messages"> = (
 			type Block = (ThinkingContent | TextContent | (ToolCall & { partialJson: string })) & { index: number };
 			const blocks = output.content as Block[];
 			stream.push({ type: "start", partial: output });
-			// Retry loop for rate-limit errors from proxies (e.g. z.ai) that the SDK doesn't handle.
-			// These errors surface when iterating the stream, so we retry the full stream creation.
-			// Only retry if no content blocks have been emitted yet (safe to restart).
+			// Retry loop for transient errors from the stream.
+			// Rate-limit/overload: only before content starts (safe to restart).
+			// Truncated JSON: also after content starts (partial response is unusable).
 			let providerRetryAttempt = 0;
 			let started = false;
 			do {
@@ -542,12 +548,15 @@ export const streamAnthropic: StreamFunction<"anthropic-messages"> = (
 					}
 					break; // Stream completed successfully
 				} catch (streamError) {
-					// Only retry if: not aborted, no content emitted yet, retries left, and error is retryable
+					// Transient stream parse errors (truncated JSON) are retryable even after content
+					// has started streaming, since the partial response is unusable anyway.
+					// Rate-limit/overload errors are only retried before content starts.
+					const isTransient = isTransientStreamParseError(streamError);
 					if (
 						options?.signal?.aborted ||
-						firstTokenTime !== undefined ||
 						providerRetryAttempt >= PROVIDER_MAX_RETRIES ||
-						!isProviderRetryableError(streamError)
+						(!isTransient && firstTokenTime !== undefined) ||
+						(!isTransient && !isProviderRetryableError(streamError))
 					) {
 						throw streamError;
 					}
@@ -557,6 +566,8 @@ export const streamAnthropic: StreamFunction<"anthropic-messages"> = (
 					// Reset output state for clean retry
 					output.content.length = 0;
 					output.stopReason = "stop";
+					firstTokenTime = undefined;
+					started = false;
 				}
 			} while (!started);
 

package/src/providers/cursor.ts

@@ -37,6 +37,7 @@ import {
 	AssistantMessageSchema,
 	BackgroundShellSpawnResultSchema,
 	ClientHeartbeatSchema,
+	ComputerUseResultSchema,
 	ConversationActionSchema,
 	type ConversationStateStructure,
 	ConversationStateStructureSchema,
@@ -70,6 +71,7 @@ import {
 	GrepUnionResultSchema,
 	KvClientMessageSchema,
 	type KvServerMessage,
+	ListMcpResourcesExecResultSchema,
 	type LsDirectoryTreeNode,
 	type LsDirectoryTreeNode_File,
 	LsDirectoryTreeNode_FileSchema,
@@ -88,9 +90,11 @@ import {
 	McpToolResultContentItemSchema,
 	ModelDetailsSchema,
 	ReadErrorSchema,
+	ReadMcpResourceExecResultSchema,
 	ReadRejectedSchema,
 	ReadResultSchema,
 	ReadSuccessSchema,
+	RecordScreenResultSchema,
 	RequestContextResultSchema,
 	RequestContextSchema,
 	RequestContextSuccessSchema,
@@ -965,6 +969,26 @@ async function handleExecServerMessage(
 			sendExecClientMessage(h2Request, execMsg, "mcpResult", execResult);
 			return;
 		}
+		case "listMcpResourcesExecArgs": {
+			const execResult = create(ListMcpResourcesExecResultSchema, {});
+			sendExecClientMessage(h2Request, execMsg, "listMcpResourcesExecResult", execResult);
+			return;
+		}
+		case "readMcpResourceExecArgs": {
+			const execResult = create(ReadMcpResourceExecResultSchema, {});
+			sendExecClientMessage(h2Request, execMsg, "readMcpResourceExecResult", execResult);
+			return;
+		}
+		case "recordScreenArgs": {
+			const execResult = create(RecordScreenResultSchema, {});
+			sendExecClientMessage(h2Request, execMsg, "recordScreenResult", execResult);
+			return;
+		}
+		case "computerUseArgs": {
+			const execResult = create(ComputerUseResultSchema, {});
+			sendExecClientMessage(h2Request, execMsg, "computerUseResult", execResult);
+			return;
+		}
 		default:
 			log("warn", "unhandledExecMessage", { execCase });
 	}