@ailib-official/ai-protocol 0.8.4 → 1.0.0

package/README.md

@@ -31,7 +31,13 @@ ai-protocol/
 │       ├── multimodal.json    # Extended multimodal capabilities schema
 │       ├── provider-contract.json # Provider contract schema
 │       ├── capability-profile.json # Experimental I/O/S phase metadata schema
-│       └── context-policy.json # Context management policy schema
+│       ├── context-policy.json # Context management policy schema
+│       ├── execution-metadata.json # Wave-5 execution metadata schema
+│       ├── metadata-model-entry.json # Typed metadata.models entries
+│       ├── message-roles.json # Message role enumeration schema
+│       ├── pack.json          # Provider pack schema
+│       ├── pricing.json       # Optional pricing schema
+│       └── tool-calling.json  # Tool calling schema
 ├── v1/                        # v1.x stable version specification
 │   ├── spec.yaml              # Basic specifications: standard parameters, event enumeration
 │   ├── providers/             # Provider configurations (split by vendor for easy PR)
@@ -44,7 +50,7 @@ ai-protocol/
 │       ├── claude.yaml        # Claude series models
 │       └── ...                # More models
 ├── v2/                        # v2 formal provider manifests
-│   └── providers/             # 6 V2 provider manifests (OpenAI, Anthropic, Google, etc.)
+│   └── providers/             # 12 V2 provider manifests (OpenAI, Anthropic, Google, DeepSeek, …)
 ├── v2-alpha/                  # v2-alpha experimental version
 │   └── spec.yaml              # Experimental operator definitions
 ├── tests/                     # Cross-runtime compliance test suite
@@ -183,6 +189,7 @@ npm run validate:schemas     # Validate JSON schema syntax only
 # Execution governance automation
 npm run drift:check          # Detect P0 provider/fixture/case drift
 npm run gate:manifest-consumption   # Cross-repo latest-manifest consumption gate
+npm run gate:manifest-authority     # Public manifest hygiene (TEST-002, no app provenance)
 npm run gate:compliance-matrix      # Cross-repo full compliance matrix gate
 npm run gate:fullchain             # One-shot governance fullchain gate
 npm run gate:fullchain:with-rollback # Fullchain + fail-fast rollback drill
@@ -196,6 +203,7 @@ node scripts/release-gate.js --report-only   # Advisory release gate report (non
 The canonical validation script is `scripts/validate.js`, which uses AJV v8 with JSON Schema 2020-12 and ajv-formats.  
 Execution governance scripts:
 - `scripts/drift-detect.js`: verifies P0 provider readiness coverage across v2 manifests, compliance fixtures, and loading cases
+- `scripts/gate-manifest-authority.js`: blocks hiddenpath URLs and application-specific provenance in manifest trees ([docs/MANIFEST_AUTHORITY.md](docs/MANIFEST_AUTHORITY.md), [ARCH-005])
 - `scripts/gate-manifest-consumption.js`: runs protocol + Rust/Python/TS latest-manifest consumption checks and writes gate report (`reports/manifest-gates/`)
 - `scripts/gate-compliance-matrix.js`: runs protocol + Rust/Python/TS compliance matrix checks and writes gate report (`reports/compliance-gates/`)
 - `scripts/gate-fullchain.js`: orchestrates drift + manifest + compliance + release gates in one run (`reports/fullchain-gates/`)

package/dist/v1/models/deepseek-chat.json

@@ -4,8 +4,8 @@
   "models": {
     "deepseek-chat": {
       "provider": "deepseek",
-      "model_id": "deepseek-chat",
-      "display_name": "DeepSeek Chat",
+      "model_id": "deepseek-v4-pro",
+      "display_name": "DeepSeek Chat (V4 Pro)",
       "verification": {
         "status": "verified",
         "verified_at": "2026-01-28",
@@ -30,8 +30,8 @@
     },
     "deepseek-reasoner": {
       "provider": "deepseek",
-      "model_id": "deepseek-reasoner",
-      "display_name": "DeepSeek Reasoner",
+      "model_id": "deepseek-v4-flash",
+      "display_name": "DeepSeek Reasoner (V4 Flash)",
       "verification": {
         "status": "verified",
         "verified_at": "2026-03-06",

package/dist/v1/models/gemini.json

@@ -1,6 +1,6 @@
 {
   "$schema": "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v1.json",
-  "protocol_version": "1.1",
+  "protocol_version": "1.5",
   "models": {
     "gemini-2.5-pro": {
       "provider": "gemini",
@@ -53,6 +53,64 @@
         "fast",
         "cost-effective"
       ]
+    },
+    "gemini-2.5-flash-lite": {
+      "provider": "gemini",
+      "model_id": "gemini-2.5-flash-lite",
+      "name": "Gemini 2.5 Flash-Lite",
+      "verification": {
+        "status": "verified",
+        "verified_at": "2026-05-17",
+        "source": "https://ai.google.dev/gemini-api/docs/models/gemini-2.5-flash-lite",
+        "notes": "Free-tier RPM/RPD vary by account and region; typical AI Studio free band is high concurrency vs Flash/Pro."
+      },
+      "context_window": 1048576,
+      "capabilities": [
+        "chat",
+        "vision",
+        "audio",
+        "video",
+        "tools",
+        "streaming",
+        "reasoning"
+      ],
+      "status": "active",
+      "tags": [
+        "google",
+        "multimodal",
+        "flash-lite",
+        "cost-effective",
+        "free-tier"
+      ]
+    },
+    "gemini-3.1-flash-lite-preview": {
+      "provider": "gemini",
+      "model_id": "gemini-3.1-flash-lite-preview",
+      "name": "Gemini 3.1 Flash-Lite (Preview)",
+      "verification": {
+        "status": "verified",
+        "verified_at": "2026-05-17",
+        "source": "https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-lite-preview",
+        "notes": "Preview model; availability and quotas may change by region."
+      },
+      "context_window": 1048576,
+      "capabilities": [
+        "chat",
+        "vision",
+        "audio",
+        "video",
+        "tools",
+        "streaming",
+        "reasoning"
+      ],
+      "status": "active",
+      "tags": [
+        "google",
+        "multimodal",
+        "flash-lite",
+        "preview",
+        "free-tier"
+      ]
     }
   }
 }

package/dist/v1/providers/gemini.json

@@ -20,7 +20,8 @@
   },
   "payload_format": "gemini_style",
   "api_families": [
-    "generate_content"
+    "generate_content",
+    "chat_completions"
   ],
   "default_api_family": "generate_content",
   "endpoints": {
@@ -28,6 +29,11 @@
       "path": "/models/{model}:generateContent",
       "method": "POST",
       "adapter": "gemini"
+    },
+    "openai_chat": {
+      "path": "/openai/chat/completions",
+      "method": "POST",
+      "adapter": "openai"
     }
   },
   "services": {
@@ -275,5 +281,9 @@
   "experimental_features": [
     "multimodal_video",
     "multimodal_audio"
+  ],
+  "notes": [
+    "OpenAI-compatible Chat Completions: POST {base_url}/openai/chat/completions with Authorization: Bearer <GEMINI_API_KEY> and OpenAI-style JSON body. See https://ai.google.dev/gemini-api/docs/openai",
+    "Native GenerateContent continues to use ?key= query authentication as configured under auth.type query_param."
   ]
 }

package/dist/v2/contracts/anthropic-messages.contract.json

@@ -0,0 +1,62 @@
+{
+  "$schema": "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v2/provider-contract.json",
+  "contract_version": "1.0",
+  "provider_id": "anthropic",
+  "api_style": "anthropic_messages",
+  "api_version": "2023-06-01",
+  "request_mapping": {
+    "message_format": "anthropic_messages",
+    "role_mapping": {
+      "system": "system",
+      "user": "user",
+      "assistant": "assistant",
+      "tool": "user"
+    },
+    "system_message_handling": "top_level_field",
+    "content_block_mapping": {
+      "text": {
+        "wrapper": "content_blocks"
+      },
+      "image": {
+        "format": "anthropic_source",
+        "base64_field": "source.data",
+        "url_field": "source.url"
+      },
+      "document": {
+        "format": "anthropic_document",
+        "type_field": "document",
+        "source_wrapper": "anthropic_source",
+        "base64_field": "source.data",
+        "mime_type_field": "source.media_type",
+        "default_mime_type": "application/pdf",
+        "ref_resolution": "error_before_encode"
+      }
+    }
+  },
+  "response_mapping": {
+    "content_path": "$.content[?(@.type=='text')].text",
+    "finish_reason_path": "$.stop_reason",
+    "usage_path": "$.usage",
+    "finish_reason_mapping": {
+      "end_turn": "stop",
+      "max_tokens": "length",
+      "tool_use": "tool_calls",
+      "stop_sequence": "stop"
+    }
+  },
+  "capability_contracts": {
+    "streaming": {
+      "protocol": "sse",
+      "supports_usage_in_stream": true,
+      "supports_tool_streaming": true,
+      "done_signal": "message_stop"
+    },
+    "tools": {
+      "definition_format": "anthropic_tools",
+      "supports_parallel": true
+    }
+  },
+  "authentication_contract": {
+    "method": "api_key_header"
+  }
+}

package/dist/v2/contracts/gemini-generate.contract.json

@@ -0,0 +1,59 @@
+{
+  "$schema": "https://raw.githubusercontent.com/ailib-official/ai-protocol/main/schemas/v2/provider-contract.json",
+  "contract_version": "1.0",
+  "provider_id": "google",
+  "api_style": "gemini_generate",
+  "api_version": "v1beta",
+  "request_mapping": {
+    "message_format": "gemini_contents",
+    "role_mapping": {
+      "system": "user",
+      "user": "user",
+      "assistant": "model",
+      "tool": "user"
+    },
+    "system_message_handling": "prepend_to_user",
+    "content_block_mapping": {
+      "text": {
+        "field": "text",
+        "wrapper": "array_of_parts"
+      },
+      "image": {
+        "format": "gemini_inline_data",
+        "base64_field": "inlineData.data",
+        "mime_type_field": "inlineData.mimeType"
+      },
+      "document": {
+        "format": "gemini_inline_data",
+        "source_wrapper": "gemini_inline_data",
+        "base64_field": "inlineData.data",
+        "mime_type_field": "inlineData.mimeType",
+        "default_mime_type": "application/pdf",
+        "ref_resolution": "error_before_encode"
+      }
+    }
+  },
+  "response_mapping": {
+    "content_path": "$.candidates[0].content.parts[?(@.text)].text",
+    "finish_reason_path": "$.candidates[0].finishReason",
+    "usage_path": "$.usageMetadata",
+    "finish_reason_mapping": {
+      "STOP": "stop",
+      "MAX_TOKENS": "length",
+      "SAFETY": "content_filter",
+      "RECITATION": "content_filter"
+    }
+  },
+  "capability_contracts": {
+    "streaming": {
+      "protocol": "sse",
+      "supports_usage_in_stream": true
+    },
+    "tools": {
+      "definition_format": "gemini_declarations"
+    }
+  },
+  "authentication_contract": {
+    "method": "query_param"
+  }
+}

package/dist/v2/providers/anthropic.json

@@ -3,7 +3,7 @@
   "id": "anthropic",
   "protocol_version": "2.0",
   "name": "Anthropic",
-  "version": "4.6.0",
+  "version": "4.8.0",
   "status": "stable",
   "category": "ai_provider",
   "official_url": "https://docs.anthropic.com",
@@ -45,6 +45,21 @@
       "overloaded_error": "overloaded"
     }
   },
+  "availability": {
+    "required": false,
+    "regions": [
+      "global"
+    ],
+    "check": {
+      "method": "GET",
+      "path": "/models",
+      "expected_status": [
+        200,
+        401
+      ],
+      "timeout_ms": 3000
+    }
+  },
   "capabilities": {
     "required": [
       "text",
@@ -67,6 +82,14 @@
       "extended_thinking": true,
       "streaming_usage": true,
       "system_messages": true
+    },
+    "tool_calling": {
+      "native": {
+        "supported": true,
+        "reliability": "full",
+        "parallel": true,
+        "streaming": true
+      }
     }
   },
   "capability_profile": {
@@ -328,23 +351,32 @@
   },
   "metadata": {
     "models": {
-      "claude-opus-4-6": {
+      "claude-opus-4-8": {
         "context_window": 1000000,
-        "max_output_tokens": 8192,
-        "release_date": "2026-02-05",
+        "max_output_tokens": 32768,
+        "release_date": "2026-06-15",
         "pricing": {
-          "input_per_1m": 5,
-          "output_per_1m": 25
+          "input_per_1m": 15,
+          "output_per_1m": 75
         }
       },
-      "claude-sonnet-4-5": {
-        "context_window": 200000,
-        "max_output_tokens": 8192,
-        "release_date": "2025-11-24",
+      "claude-sonnet-4-6": {
+        "context_window": 1000000,
+        "max_output_tokens": 32768,
+        "release_date": "2026-06-11",
         "pricing": {
           "input_per_1m": 3,
           "output_per_1m": 15
         }
+      },
+      "claude-haiku-4-5": {
+        "context_window": 200000,
+        "max_output_tokens": 64000,
+        "release_date": "2025-10-01",
+        "pricing": {
+          "input_per_1m": 0.8,
+          "output_per_1m": 4
+        }
       }
     },
     "context_management": {

package/dist/v2/providers/cohere.json

@@ -31,6 +31,21 @@
       "503": "overloaded"
     }
   },
+  "availability": {
+    "required": false,
+    "regions": [
+      "global"
+    ],
+    "check": {
+      "method": "GET",
+      "path": "/models",
+      "expected_status": [
+        200,
+        401
+      ],
+      "timeout_ms": 3000
+    }
+  },
   "capabilities": {
     "required": [
       "text",
@@ -44,6 +59,12 @@
     "feature_flags": {
       "structured_output": true,
       "system_messages": true
+    },
+    "tool_calling": {
+      "native": {
+        "supported": true,
+        "reliability": "full"
+      }
     }
   },
   "capability_profile": {

package/dist/v2/providers/deepseek.json

@@ -3,13 +3,13 @@
   "id": "deepseek",
   "protocol_version": "2.0",
   "name": "DeepSeek",
-  "version": "3.2.0",
+  "version": "4.0.0",
   "status": "stable",
   "category": "ai_provider",
   "official_url": "https://api-docs.deepseek.com/",
   "support_contact": "https://api-docs.deepseek.com/",
   "endpoint": {
-    "base_url": "https://api.deepseek.com/v1",
+    "base_url": "https://api.deepseek.com",
     "chat": "/chat/completions",
     "auth": {
       "type": "bearer",
@@ -22,6 +22,7 @@
     "by_http_status": {
       "400": "invalid_request",
       "401": "authentication",
+      "402": "insufficient_quota",
       "403": "permission_denied",
       "404": "not_found",
       "413": "request_too_large",
@@ -36,6 +37,22 @@
       "insufficient_quota": "quota_exhausted"
     }
   },
+  "availability": {
+    "required": false,
+    "regions": [
+      "cn",
+      "global"
+    ],
+    "check": {
+      "method": "GET",
+      "path": "/models",
+      "expected_status": [
+        200,
+        401
+      ],
+      "timeout_ms": 3000
+    }
+  },
   "capabilities": {
     "required": [
       "text",
@@ -43,7 +60,6 @@
       "tools"
     ],
     "optional": [
-      "vision",
       "parallel_tools",
       "agentic",
       "reasoning",
@@ -56,6 +72,37 @@
       "extended_thinking": true,
       "streaming_usage": true,
       "system_messages": true
+    },
+    "tool_calling": {
+      "native": {
+        "supported": true,
+        "reliability": "partial",
+        "parallel": true,
+        "streaming": true,
+        "notes": "DeepSeek native tool calling is unreliable — LLM often outputs <shell> instead; text fallback required"
+      },
+      "text_fallback": {
+        "format": "xml_json",
+        "wrapper": "tool_call",
+        "body": "json",
+        "name_location": "json_field",
+        "args_key": "arguments",
+        "known_dialects": [
+          {
+            "tag": "shell",
+            "map_to": "shell"
+          },
+          {
+            "tag": "bash",
+            "map_to": "shell"
+          },
+          {
+            "tag": "function",
+            "map_to": ""
+          }
+        ],
+        "prompt_level": "L2"
+      }
     }
   },
   "capability_profile": {
@@ -74,8 +121,7 @@
     },
     "systems": {
       "requires": [
-        "mcp",
-        "search"
+        "mcp"
       ]
     }
   },
@@ -116,16 +162,17 @@
     },
     "stream": {
       "type": "boolean"
+    },
+    "reasoning_effort": {
+      "type": "string",
+      "enum": [
+        "low",
+        "medium",
+        "high"
+      ],
+      "default": "medium"
     }
   },
-  "parameter_mappings": {
-    "temperature": "temperature",
-    "max_tokens": "max_tokens",
-    "stream": "stream",
-    "top_p": "top_p",
-    "tools": "tools",
-    "tool_choice": "tool_choice"
-  },
   "streaming": {
     "decoder": {
       "format": "sse",
@@ -144,6 +191,13 @@
           "content": "$.choices[*].delta.content"
         }
       },
+      {
+        "match": "exists($.choices[*].delta.reasoning_content)",
+        "emit": "ThinkingDelta",
+        "fields": {
+          "thinking": "$.choices[*].delta.reasoning_content"
+        }
+      },
       {
         "match": "exists($.choices[*].delta.tool_calls)",
         "emit": "PartialToolCall",
@@ -171,18 +225,7 @@
   "multimodal": {
     "input": {
       "vision": {
-        "supported": true,
-        "formats": [
-          "jpeg",
-          "png",
-          "gif",
-          "webp"
-        ],
-        "encoding_methods": [
-          "base64_inline",
-          "url"
-        ],
-        "document_understanding": true
+        "supported": false
       },
       "audio": {
         "supported": false
@@ -202,20 +245,26 @@
     }
   },
   "api_families": [
-    "chat_completions"
+    "chat_completions",
+    "anthropic_messages"
   ],
   "default_api_family": "chat_completions",
   "endpoints": {
-    "chat": {
+    "chat_openai": {
       "path": "/chat/completions",
       "method": "POST",
       "adapter": "openai"
+    },
+    "chat_anthropic": {
+      "path": "/anthropic/v1/messages",
+      "method": "POST",
+      "adapter": "anthropic"
     }
   },
   "retry_policy": {
     "strategy": "exponential_backoff",
     "max_retries": 3,
-    "min_delay_ms": 1000,
+    "min_delay_ms": 500,
     "max_delay_ms": 30000,
     "jitter": "full",
     "retry_on_http_status": [
@@ -234,37 +283,62 @@
     }
   },
   "metadata": {
-    "api_compatibility": "openai",
+    "api_compatibility": [
+      "openai",
+      "anthropic"
+    ],
     "open_source": {
-      "license": "MIT",
-      "hugging_face": "https://huggingface.co/deepseek-ai/DeepSeek-V3.2"
+      "license": "MIT (weights)",
+      "hugging_face": "https://huggingface.co/collections/deepseek-ai/deepseek-v4"
     },
     "models": {
-      "deepseek-chat": {
-        "context_window": 128000,
-        "max_output_tokens": 8192,
+      "deepseek-v4-pro": {
+        "context_window": 1048576,
+        "max_output_tokens": 393216,
+        "release_date": "2026-04-24",
         "architecture": {
           "type": "moe",
-          "total": "671B",
-          "active": "37B",
-          "experts": 256
+          "total": "1.6T",
+          "active": "49B"
         },
         "pricing": {
-          "input_per_1m": 0.28,
-          "output_per_1m": 0.42
-        }
+          "input_per_1m": 0.435,
+          "output_per_1m": 0.87,
+          "cache_hit_input_per_1m": 0.003625
+        },
+        "thinking": "dual_mode"
       },
-      "deepseek-reasoner": {
-        "context_window": 128000,
-        "max_output_tokens": 8192,
+      "deepseek-v4-flash": {
+        "context_window": 1048576,
+        "max_output_tokens": 393216,
+        "release_date": "2026-04-24",
+        "architecture": {
+          "type": "moe",
+          "total": "284B",
+          "active": "13B"
+        },
         "pricing": {
-          "input_per_1m": 0.55,
-          "output_per_1m": 2.19
-        }
+          "input_per_1m": 0.14,
+          "output_per_1m": 0.28,
+          "cache_hit_input_per_1m": 0.0028
+        },
+        "thinking": "dual_mode"
+      }
+    },
+    "deprecated": {
+      "deepseek-chat": {
+        "retired_on": "2026-07-24",
+        "maps_to": "deepseek-v4-flash (non-thinking)",
+        "note": "Use deepseek-v4-flash instead"
+      },
+      "deepseek-reasoner": {
+        "retired_on": "2026-07-24",
+        "maps_to": "deepseek-v4-flash (thinking mode)",
+        "note": "Use deepseek-v4-flash with thinking mode instead"
       }
     },
     "sdk": {
-      "note": "OpenAI-compatible — use openai SDK with base_url override",
+      "note": "OpenAI-compatible — use openai SDK with base_url override. Also supports Anthropic Messages API format.",
       "python": "openai",
       "typescript": "openai"
     }