User Guide

1. Pick a workspace (group + doc type) from the sidebar dropdowns.
2. Drag a PDF into the upload zone on the Upload tab, or click to browse.
3. Click Ingest (Fast) to convert it to Markdown immediately, or Ingest (Standard) for background processing with full DOCINDEX tree.
4. Monitor progress in the server log panel at the bottom.

1. Switch to the Archive tab — ingested files appear in the list.
2. Click Preview to see what dimensions the LLM inferred (asset manager, fund name, reporting date, etc.).
3. Confirm or edit the values, then click Archive to move the file to the permanent tree.
4. The document index (_index/index.json) is updated automatically.

1. Switch to the Extract tab.
2. Select the schema to use (from _extract-templates/ in your workspace).
3. Pick the archived document(s) to extract from.
4. Click Run Extract — the LLM populates each field in the schema.
5. Results are saved as {stem}-singles.md and {stem}-tables.md alongside the PDF.

1. Switch to the Query tab.
2. Use the filter panel to select a date range and dimensions.
3. Click Snapshot for point-in-time data or Series for time-series across reports.
4. Use the Chat sub-tab to ask natural-language questions — answers are grounded in the archived documents.

Two auth paths. Use workspace auth for programmatic API access (curl/SDK). Use user login when calling from a browser context or testing user flows.

Workspace auth (master password)

TOKEN=$(curl -s -X POST http://localhost:8000/auth/login \
  -d "password=YOUR_MASTER_PASSWORD" \
  | python3 -c "import sys,json; print(json.load(sys.stdin)['access_token'])")

User login (group password)

TOKEN=$(curl -s -X POST http://localhost:8000/auth/user/login \
  -H "Content-Type: application/json" \
  -d '{"username":"reedcapital","password":"Reed123!","group_id":1}' \
  | python3 -c "import sys,json; print(json.load(sys.stdin)['access_token'])")

All subsequent requests: -H "Authorization: Bearer $TOKEN"

curl -X POST "http://localhost:8000/api/fin-report/ingest-fast" \
  -H "Authorization: Bearer $TOKEN" \
  -F "files=@/path/to/report.pdf"

Use /ingest for background DOCINDEX processing. Use /ingest-fast for synchronous fast Markdown output.

Archives every ingested file in the temp folder. The LLM infers dimension values automatically.

curl -X POST "http://localhost:8000/api/fin-report/archive" \
  -H "Authorization: Bearer $TOKEN"

# Single document
curl -X POST "http://localhost:8000/api/fin-report/extract" \
  -H "Authorization: Bearer $TOKEN" \
  -G \
  --data-urlencode "path_level2=amundi_initiative-impact" \
  --data-urlencode "path_level3=2024q4" \
  --data-urlencode "schema_suffix=prtf"

# Entire snapshot date (all entities)
curl -X POST "http://localhost:8000/api/fin-report/extract/batch" \
  -H "Authorization: Bearer $TOKEN" \
  -G \
  --data-urlencode "path_level3=2024q4" \
  --data-urlencode "schema_suffix=prtf"

# Snapshot — latest document per entity at a given date
curl -X POST "http://localhost:8000/api/fin-report/select/snapshot?date=2024q4" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{}'

# Series — all documents across a date range
curl -X POST "http://localhost:8000/api/fin-report/select/series?start_date=2024q1&end_date=2024q4" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{}'

# With dimension filters
curl -X POST "http://localhost:8000/api/fin-report/select/snapshot?date=2024q4" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"config_params": {"asset-manager": "amundi"}}'

Response: {"matched_files": ["reedcapital/fin-report/amundi.../2024q4/..."], "count": 3}

Pass matched_files from a select call to ground the LLM answer in real documents.

curl -X POST "http://localhost:8000/api/fin-report/chat" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "matched_files": [
      "reedcapital/fin-report/amundi_initiative-impact/2024q4/amundi_initiative-impact_2024q4"
    ],
    "prompt": "What is the total AUM as of Q4 2024?"
  }'