{
  "version": "0.2",
  "semantic": {
    "language": "en",
    "title": "Dify · From documents to answers",
    "subtitle": "1.17.0 · Three layers · Knowledge ingestion and basic Chat",
    "summary": "Blue: document ingestion. Green: online chat. Gray: shared interfaces. Arrows show selected calls or task delivery, not timing, all dependencies, or one global execution order.",
    "intent": {
      "diagram_kind": "architecture",
      "primary_question": "Which services handle indexing and online chat, and what does each read or write?",
      "audience": [
        "Developers new to Dify architecture"
      ],
      "scope": "Self-hosted Dify 1.17.0; local text uploads; high_quality paragraph indexing; basic AppMode.CHAT; a configured and selected internal dataset with semantic_search; successful paths.",
      "exclusions": [
        "Not all of Dify: Workflow, Chatflow, Agent, knowledge pipelines, multimodal processing, reranking, errors and retries are excluded.",
        "Three logical layers are not three deployed services. Chat and RAG are internal API code, not separate microservices. Other app modes may use asynchronous queues.",
        "Model calls are traced only to the Plugin Daemon HTTP interface. Plugin execution and downstream providers were not reviewed, so downstream edges are not drawn.",
        "Other SQL/Redis consumers, Web server hosting, authentication, callback details and cache-hit branches are omitted from the overview.",
        "Dify has not been run and no real model calls were made. Independent semantic review is pending."
      ]
    },
    "entities": [
      {
        "id": "web",
        "label": "Web console",
        "description": "Browser client\nUpload · create",
        "kind": "browser",
        "role": "process",
        "importance": "primary",
        "source_refs": [
          "src-web",
          "src-file"
        ]
      },
      {
        "id": "nginx",
        "label": "Nginx",
        "description": "Reverse proxy\nURL routing",
        "kind": "gateway",
        "role": "process",
        "importance": "primary",
        "source_refs": [
          "src-nginx"
        ]
      },
      {
        "id": "caller",
        "label": "API client",
        "description": "Chat requests\nStreamed replies",
        "kind": "client",
        "role": "process",
        "importance": "primary",
        "source_refs": [
          "src-chat-api",
          "src-nginx"
        ]
      },
      {
        "id": "api",
        "label": "API service",
        "description": "Chat · RAG\nIn-process thread",
        "kind": "server",
        "role": "process",
        "importance": "primary",
        "source_refs": [
          "src-deploy",
          "src-dispatch",
          "src-thread",
          "src-chat",
          "src-file"
        ]
      },
      {
        "id": "worker",
        "label": "Indexer",
        "description": "Celery worker\nExtract · split",
        "kind": "worker",
        "role": "process",
        "importance": "primary",
        "source_refs": [
          "src-deploy",
          "src-task",
          "src-index"
        ]
      },
      {
        "id": "plugin",
        "label": "Plugins",
        "description": "Plugin Daemon\nModel boundary",
        "kind": "plugin",
        "role": "process",
        "importance": "primary",
        "source_refs": [
          "src-plugin-deploy",
          "src-plugin-llm",
          "src-plugin-embedding",
          "src-plugin-base"
        ]
      },
      {
        "id": "files",
        "label": "File store",
        "description": "Original files\nLocal by default",
        "kind": "storage",
        "role": "process",
        "importance": "primary",
        "source_refs": [
          "src-file",
          "src-extract",
          "src-deploy",
          "src-stores"
        ]
      },
      {
        "id": "redis",
        "label": "Redis",
        "description": "Task broker\nQuery vector cache",
        "kind": "queue",
        "role": "process",
        "importance": "primary",
        "source_refs": [
          "src-broker",
          "src-broker-default",
          "src-query-embedding"
        ]
      },
      {
        "id": "sql",
        "label": "SQL store",
        "description": "Metadata · chunks\nPostgreSQL here",
        "kind": "database",
        "role": "process",
        "importance": "primary",
        "source_refs": [
          "src-file",
          "src-index",
          "src-postgres"
        ]
      },
      {
        "id": "vector",
        "label": "Vector DB",
        "description": "Document vectors\nWeaviate here",
        "kind": "database",
        "role": "process",
        "importance": "primary",
        "source_refs": [
          "src-vector-create",
          "src-vector-query",
          "src-stores"
        ]
      }
    ],
    "groups": [
      {
        "id": "access",
        "label": "01  Access · Requests",
        "kind": "logical-layer",
        "members": [
          "web",
          "nginx",
          "caller"
        ]
      },
      {
        "id": "services",
        "label": "02  Execution · API and indexing",
        "kind": "logical-layer",
        "members": [
          "api",
          "worker",
          "plugin"
        ]
      },
      {
        "id": "data",
        "label": "03  Data · Files, tasks and vectors",
        "kind": "logical-layer",
        "members": [
          "files",
          "redis",
          "sql",
          "vector"
        ]
      }
    ],
    "relations": [
      {
        "id": "web-nginx",
        "from": "web",
        "to": "nginx",
        "label": "Upload / create",
        "kind": "http",
        "direction": "forward",
        "condition": "Console request succeeds",
        "importance": "primary",
        "source_refs": [
          "src-web",
          "src-file",
          "src-nginx"
        ]
      },
      {
        "id": "caller-nginx",
        "from": "caller",
        "to": "nginx",
        "label": "Ask / SSE",
        "kind": "http",
        "direction": "bidirectional",
        "condition": "Basic CHAT; streaming response mode",
        "importance": "primary",
        "source_refs": [
          "src-nginx",
          "src-chat-api"
        ]
      },
      {
        "id": "nginx-api",
        "from": "nginx",
        "to": "api",
        "label": "API request / reply",
        "kind": "http",
        "direction": "bidirectional",
        "condition": "Console API or /v1 route",
        "importance": "primary",
        "source_refs": [
          "src-nginx",
          "src-chat-api"
        ]
      },
      {
        "id": "api-files",
        "from": "api",
        "to": "files",
        "label": "Store file",
        "kind": "storage-access",
        "direction": "forward",
        "condition": "Local uploaded text file; validation succeeds",
        "importance": "primary",
        "source_refs": [
          "src-file"
        ]
      },
      {
        "id": "api-sql",
        "from": "api",
        "to": "sql",
        "label": "Store metadata",
        "kind": "database-access",
        "direction": "forward",
        "condition": "Upload and document creation succeed",
        "importance": "primary",
        "source_refs": [
          "src-file",
          "src-document"
        ]
      },
      {
        "id": "api-redis",
        "from": "api",
        "to": "redis",
        "label": "Tasks / query cache",
        "kind": "dependency",
        "direction": "forward",
        "condition": "Index dispatch after document commit; query embedding cache access on semantic retrieval",
        "importance": "primary",
        "source_refs": [
          "src-document",
          "src-task-proxy",
          "src-task-policy",
          "src-broker",
          "src-broker-default",
          "src-query-embedding"
        ]
      },
      {
        "id": "redis-worker",
        "from": "redis",
        "to": "worker",
        "label": "Consume index job",
        "kind": "task-delivery",
        "direction": "forward",
        "condition": "Worker consumes Celery broker; self-hosted priority_dataset queue",
        "importance": "primary",
        "source_refs": [
          "src-broker",
          "src-task-policy",
          "src-task",
          "src-task-run"
        ]
      },
      {
        "id": "worker-files",
        "from": "worker",
        "to": "files",
        "label": "Read source file",
        "kind": "storage-access",
        "direction": "forward",
        "condition": "FILE source in extraction",
        "importance": "primary",
        "source_refs": [
          "src-task-run",
          "src-index",
          "src-extract"
        ]
      },
      {
        "id": "worker-sql",
        "from": "worker",
        "to": "sql",
        "label": "Store chunks",
        "kind": "database-access",
        "direction": "forward",
        "condition": "Successful paragraph indexing",
        "importance": "primary",
        "source_refs": [
          "src-index",
          "src-doc-embedding"
        ]
      },
      {
        "id": "worker-vector",
        "from": "worker",
        "to": "vector",
        "label": "Write index",
        "kind": "vector-access",
        "direction": "forward",
        "condition": "high_quality paragraph indexing; embedding succeeds",
        "importance": "primary",
        "source_refs": [
          "src-index",
          "src-paragraph",
          "src-vector-create"
        ]
      },
      {
        "id": "api-vector",
        "from": "api",
        "to": "vector",
        "label": "Semantic search",
        "kind": "vector-access",
        "direction": "forward",
        "condition": "Dataset configured and selected; internal high_quality semantic retrieval; filters permit results",
        "importance": "primary",
        "source_refs": [
          "src-chat",
          "src-retrieval",
          "src-search",
          "src-vector-query"
        ]
      },
      {
        "id": "worker-plugin",
        "from": "worker",
        "to": "plugin",
        "label": "Embed documents",
        "kind": "model-invocation",
        "direction": "forward",
        "condition": "Document embedding cache miss; model configured",
        "importance": "primary",
        "source_refs": [
          "src-vector-create",
          "src-doc-embedding",
          "src-model",
          "src-plugin-embedding",
          "src-plugin-base"
        ]
      },
      {
        "id": "api-plugin",
        "from": "api",
        "to": "plugin",
        "label": "Embedding / LLM",
        "kind": "model-invocation",
        "direction": "forward",
        "condition": "Query embedding cache miss and/or LLM invocation; validated normal CHAT path",
        "importance": "primary",
        "source_refs": [
          "src-chat",
          "src-query-embedding",
          "src-model",
          "src-plugin-runtime",
          "src-plugin-llm",
          "src-plugin-embedding",
          "src-plugin-base"
        ]
      }
    ],
    "flows": [
      {
        "id": "index-path",
        "label": "Index dispatch and write",
        "importance": "primary",
        "repeat": "event-driven",
        "relation_ids": [
          "web-nginx",
          "nginx-api",
          "api-redis",
          "redis-worker",
          "worker-vector"
        ]
      },
      {
        "id": "retrieval-path",
        "label": "Retrieval within chat",
        "importance": "primary",
        "repeat": "event-driven",
        "relation_ids": [
          "caller-nginx",
          "nginx-api",
          "api-vector"
        ]
      }
    ],
    "sources": [
      {
        "id": "src-web",
        "type": "repository",
        "title": "Web creates dataset documents",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "web/service/datasets.ts",
          "line": 153,
          "end_line": 170
        }
      },
      {
        "id": "src-nginx",
        "type": "repository",
        "title": "Nginx API routes",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "docker/nginx/conf.d/default.conf.template",
          "line": 8,
          "end_line": 45
        }
      },
      {
        "id": "src-deploy",
        "type": "repository",
        "title": "API and Worker are separate services",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "docker/docker-compose.yaml",
          "line": 227,
          "end_line": 352
        }
      },
      {
        "id": "src-plugin-deploy",
        "type": "repository",
        "title": "Plugin daemon service",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "docker/docker-compose.yaml",
          "line": 573,
          "end_line": 602
        }
      },
      {
        "id": "src-file",
        "type": "repository",
        "title": "File bytes and SQL metadata are separate",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/services/file_service.py",
          "line": 37,
          "end_line": 125
        }
      },
      {
        "id": "src-document",
        "type": "repository",
        "title": "Commit documents before submitting indexing",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/services/dataset_service.py",
          "line": 2482,
          "end_line": 2499
        }
      },
      {
        "id": "src-broker",
        "type": "repository",
        "title": "Celery broker configuration",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/extensions/ext_celery.py",
          "line": 113,
          "end_line": 132
        }
      },
      {
        "id": "src-broker-default",
        "type": "repository",
        "title": "Redis broker default",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "docker/.env.example",
          "line": 115,
          "end_line": 124
        }
      },
      {
        "id": "src-stores",
        "type": "repository",
        "title": "Storage and vector store defaults",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "docker/.env.example",
          "line": 159,
          "end_line": 164
        }
      },
      {
        "id": "src-postgres",
        "type": "repository",
        "title": "PostgreSQL deployment option",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "docker/docker-compose.yaml",
          "line": 425,
          "end_line": 437
        }
      },
      {
        "id": "src-task-proxy",
        "type": "repository",
        "title": "Direct Celery dispatch",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/services/document_indexing_proxy/batch_indexing_base.py",
          "line": 37,
          "end_line": 47
        }
      },
      {
        "id": "src-task-policy",
        "type": "repository",
        "title": "Self-hosted dispatch policy",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/services/document_indexing_proxy/base.py",
          "line": 79,
          "end_line": 111
        }
      },
      {
        "id": "src-task",
        "type": "repository",
        "title": "Indexing task entry points",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/tasks/document_indexing_task.py",
          "line": 200,
          "end_line": 267
        }
      },
      {
        "id": "src-task-run",
        "type": "repository",
        "title": "Task invokes IndexingRunner",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/tasks/document_indexing_task.py",
          "line": 49,
          "end_line": 125
        }
      },
      {
        "id": "src-index",
        "type": "repository",
        "title": "Extract, transform, persist segments, load index",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/indexing_runner.py",
          "line": 81,
          "end_line": 159
        }
      },
      {
        "id": "src-extract",
        "type": "repository",
        "title": "Download uploaded file for extraction",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/rag/extractor/extract_processor.py",
          "line": 115,
          "end_line": 137
        }
      },
      {
        "id": "src-paragraph",
        "type": "repository",
        "title": "High-quality paragraph vector indexing",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/rag/index_processor/processor/paragraph_index_processor.py",
          "line": 134,
          "end_line": 149
        }
      },
      {
        "id": "src-vector-create",
        "type": "repository",
        "title": "Embed documents then write vectors",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/rag/datasource/vdb/vector_factory.py",
          "line": 168,
          "end_line": 189
        }
      },
      {
        "id": "src-vector-query",
        "type": "repository",
        "title": "Embed query then search vectors",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/rag/datasource/vdb/vector_factory.py",
          "line": 252,
          "end_line": 259
        }
      },
      {
        "id": "src-doc-embedding",
        "type": "repository",
        "title": "Document embedding SQL cache",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/rag/embedding/cached_embedding.py",
          "line": 29,
          "end_line": 103
        }
      },
      {
        "id": "src-query-embedding",
        "type": "repository",
        "title": "Query embedding Redis cache",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/rag/embedding/cached_embedding.py",
          "line": 194,
          "end_line": 241
        }
      },
      {
        "id": "src-chat-api",
        "type": "repository",
        "title": "Generate chat and return response stream",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/controllers/service_api/app/completion.py",
          "line": 388,
          "end_line": 419
        }
      },
      {
        "id": "src-dispatch",
        "type": "repository",
        "title": "Basic CHAT dispatch",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/services/app_generate_service.py",
          "line": 218,
          "end_line": 231
        }
      },
      {
        "id": "src-thread",
        "type": "repository",
        "title": "API-local thread and ChatAppRunner",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/app/apps/chat/app_generator.py",
          "line": 208,
          "end_line": 272
        }
      },
      {
        "id": "src-chat",
        "type": "repository",
        "title": "Retrieve context then generate answer",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/app/apps/chat/app_runner.py",
          "line": 160,
          "end_line": 256
        }
      },
      {
        "id": "src-retrieval",
        "type": "repository",
        "title": "Selected internal dataset retrieval",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/rag/retrieval/dataset_retrieval.py",
          "line": 660,
          "end_line": 754
        }
      },
      {
        "id": "src-search",
        "type": "repository",
        "title": "Semantic vector retrieval",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/rag/datasource/retrieval_service.py",
          "line": 303,
          "end_line": 347
        }
      },
      {
        "id": "src-model",
        "type": "repository",
        "title": "LLM and embedding model invocation",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/model_manager.py",
          "line": 181,
          "end_line": 266
        }
      },
      {
        "id": "src-plugin-runtime",
        "type": "repository",
        "title": "Model runtime delegates to plugin client",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/plugin/impl/model_runtime.py",
          "line": 309,
          "end_line": 351
        }
      },
      {
        "id": "src-plugin-llm",
        "type": "repository",
        "title": "Plugin daemon LLM HTTP contract",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/plugin/impl/model.py",
          "line": 170,
          "end_line": 217
        }
      },
      {
        "id": "src-plugin-embedding",
        "type": "repository",
        "title": "Plugin daemon embedding HTTP contract",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/plugin/impl/model.py",
          "line": 360,
          "end_line": 399
        }
      },
      {
        "id": "src-plugin-base",
        "type": "repository",
        "title": "HTTP client points at plugin daemon",
        "repository": {
          "url": "https://github.com/langgenius/dify",
          "revision": "09a855dcef24c0edc7431c46c0cfaa494481daf5",
          "path": "api/core/plugin/impl/base.py",
          "line": 98,
          "end_line": 165
        }
      }
    ]
  },
  "presentation": {
    "icon_system": "auto",
    "style": "deep-tech",
    "layout": "layered",
    "motion": "showcase-v1"
  },
  "presentation_sources": {
    "style": "explicit",
    "layout": "model"
  },
  "reader": {
    "enabled": true,
    "views": [
      {
        "id": "ingest",
        "label": "Document ingestion",
        "relation_ids": [
          "web-nginx",
          "nginx-api",
          "api-files",
          "api-sql",
          "api-redis",
          "redis-worker",
          "worker-files",
          "worker-sql",
          "worker-vector",
          "worker-plugin"
        ]
      },
      {
        "id": "query",
        "label": "Online chat",
        "relation_ids": [
          "caller-nginx",
          "nginx-api",
          "api-redis",
          "api-vector",
          "api-plugin"
        ]
      }
    ]
  }
}