How narenjcs/assignment works
DocIntel is a cross-cloud document-intelligence application. Users upload a PDF or DOCX in the React UI, then follow job status and results or chat about a selected job. The AWS API and upload trigger create and dispatch jobs; an AgentCore orchestrator selects the DOCX or PDF processing path. AWS and Databricks MCP tools bridge document access and result persistence across clouds. The README documents both sync and async processing and result storage in DynamoDB and Unity Catalog. Several connections below rely on that documentation because the corresponding source was not excerpted.
Main entry points: server.py Databricks MCP server · App.tsx React workspace · main.py DOCX agent · handler.ts AWS MCP tools
Components 26
Browser experience
- React workspace
frontend/src/app/App.tsx - Upload flow
frontend/src/features/upload/useUpload.ts - Jobs and results
frontend/src/features/jobs/JobsList.tsx - Job chat
frontend/src/features/chat/useChatStream.ts - API client
frontend/src/lib/api.ts
AWS application
- HTTP API
aws/lambdas/src/api/handler.ts - Upload, jobs, chat
aws/lambdas/src/api/routes/uploads.ts - Upload storage
aws/infra/lib/constructs/storage.ts - Upload trigger
aws/lambdas/src/s3-trigger/handler.ts - Job state
aws/lambdas/src/lib/jobs.ts - AWS MCP tools
aws/lambdas/src/mcp-tools/handler.ts - AgentCore Gateway
aws/common/docintel_common/mcp_backend.py - Shared runtime config
aws/common/docintel_common/config.py
Agent processing
- Workflow orchestrator
aws/agents/orchestrator/main.py - Routing and workflow
aws/agents/orchestrator/workflow.py - DOCX agent
aws/agents/docx_agent/main.py
Databricks processing
- Databricks MCP server
databricks/app/src/docintel_app/server.py - PDF agent
databricks/app/src/docintel_app/agent.py - PDF ingestion
databricks/app/src/docintel_app/tools/ingest.py - PDF extraction
databricks/app/src/docintel_app/tools/extract.py - Result enrichment
databricks/app/src/docintel_app/tools/enrich.py - Result persistence
databricks/app/src/docintel_app/tools/persist.py - Unity Catalog
databricks/app/src/docintel_app/uc.py
Other components
- User
- Databricks services
- Bedrock model
Connections 37
- User → React workspace: uses
README.md - React workspace → Upload flow: shows
frontend/src/app/App.tsx - React workspace → Jobs and results: shows
frontend/src/app/App.tsx - React workspace → Job chat: shows
frontend/src/app/App.tsx - Upload flow → API client: calls
frontend/src/features/upload/useUpload.ts - Jobs and results → API client: polls
- Job chat → API client: streams
frontend/src/features/chat/useChatStream.ts - API client → HTTP API: requests
README.md - HTTP API → Upload, jobs, chat: dispatches
aws/lambdas/src/api/handler.ts - Upload, jobs, chat → Upload storage: creates upload
README.md - Upload storage → Upload trigger: notifies
README.md - Upload trigger → Job state: updates status
aws/lambdas/src/s3-trigger/handler.ts - Upload trigger → Workflow orchestrator: invokes async
aws/lambdas/src/s3-trigger/handler.ts - HTTP API → Workflow orchestrator: starts processing
README.md - Workflow orchestrator → Routing and workflow: runs
aws/agents/orchestrator/main.py - Routing and workflow → DOCX agent: routes DOCX
README.md - Routing and workflow → Databricks MCP server: routes PDF
README.md - DOCX agent → AgentCore Gateway: calls tools
aws/agents/docx_agent/main.py - AgentCore Gateway → AWS MCP tools: connects tools
README.md - AWS MCP tools → Upload storage: reads document
README.md - AWS MCP tools → Job state: writes results
README.md - Databricks MCP server → PDF agent: runs agent
databricks/app/src/docintel_app/server.py - Databricks MCP server → PDF ingestion: registers tool
databricks/app/src/docintel_app/server.py - Databricks MCP server → PDF extraction: registers tool
databricks/app/src/docintel_app/server.py - Databricks MCP server → Result enrichment: registers tool
databricks/app/src/docintel_app/server.py - Databricks MCP server → Result persistence: registers tool
databricks/app/src/docintel_app/server.py - PDF ingestion → AWS MCP tools: fetches document
databricks/app/src/docintel_app/tools/ingest.py - PDF ingestion → Unity Catalog: uploads PDF
databricks/app/src/docintel_app/tools/ingest.py - PDF extraction → Databricks services: uses PDF services
README.md - Result enrichment → Databricks services: uses model service
README.md - Result persistence → Unity Catalog: writes result
README.md - PDF agent → Bedrock model: uses model
README.md - DOCX agent → Bedrock model: uses model
aws/agents/docx_agent/main.py - Databricks MCP server → Databricks services: serves MCP
databricks/app/src/docintel_app/server.py - Workflow orchestrator → Job state: tracks job
README.md - DOCX agent → Job state: saves result
README.md - Shared runtime config → AgentCore Gateway: configures client
aws/agents/docx_agent/main.py