A lightweight, modular marketplace platform that mounts full-blown FastAPI applications and provides an AI Agent-Friendly Financial & Business Intelligence Gateway.
Humans enter financial and inventory data via streamlined web interfaces or statement imports, while autonomous AI agents and pair-programming LLMs extract instant, token-optimized executive summaries.
Run the platform (hardcoded to port 11173):
uv run python main.py- Marketplace Hub: http://localhost:11173
- AI Agent Intelligence Hub: http://localhost:11173/agent
- Agent Discovery (llms.txt): http://localhost:11173/llms.txt
- Core API Documentation: http://localhost:11173/docs
The platform exposes dedicated machine-readable endpoints designed for sub-5ms response times and minimal token usage:
| Endpoint | Method | Format | Description |
|---|---|---|---|
/agent/summary |
GET |
JSON | Unified Executive Brief: Combines Personal Finance (cashflow, envelope budgets) + Business Performance (revenue, COGS, margins, receivables, net worth). |
/agent/brief |
GET |
Markdown | Token-Optimized Markdown: ~230 tokens, formatted for direct injection into LLM prompts or chat responses. |
/agent/personal-finance |
GET |
JSON / MD | Personal finance envelopes, cashflow, top expenses, and uncategorized transactions. Supports ?month=X&year=Y. |
/agent/business-performance |
GET |
JSON / MD | Business sales, COGS, gross margin, cash collected, debtors, inventory valuation, and conservative/optimistic net worth. |
/agent/query?q=<term> |
GET |
JSON | Semantic parameter resolver for targeted lookups (net_worth, revenue, debtors, budget, inventory, alerts). |
/agent/tools.json |
GET |
JSON | Pre-built function calling schemas (OpenAI / Anthropic / Gemini format). |
/llms.txt |
GET |
Plain Text | Standard llms.txt specification index for agent discovery. |
/llms-full.txt |
GET |
Plain Text | Comprehensive agent documentation and endpoint schema guide. |
Humans can input data through multiple convenient interfaces:
- Quick Single Entry: Add individual transactions on the fly via
/transactionsor the/agenthub. - Statement CSV Upload: Drop raw bank/UPI statements into
/personal-finance/upload(automatically strips statement noise and deduplicates). - Envelope Budgeting: Configure category monthly limits at
/personal-finance/settings. - Categorization: Classify transactions with forward-matching rules.
- Inventory Batches: Group inventory by supplier at
/asset-manager/batch. - Item Cataloguing: Log purchase price, intended markup/profit %, and serial numbers.
- Sales & Customer Ledger: Record sales, track credit, and log installment payments at
/asset-manager/salesand/asset-manager/ledger. - Fixed Assets: Track equipment and workshop tooling at
/asset-manager/assets-list.
Agents can import the function schema directly from /agent/tools.json or call the REST APIs directly:
curl http://localhost:11173/agent/summarycurl http://localhost:11173/agent/briefcurl -X POST http://localhost:11173/personal-finance/api/agent/transactions \
-H "Content-Type: application/json" \
-d '{
"tran_date": "2026-11-09",
"particulars": "Stripe Merchant Payout",
"credit": 1250.00,
"category": "Income"
}'curl -X POST http://localhost:11173/asset-manager/api/agent/payments \
-H "Content-Type: application/json" \
-d '{
"sale_id": 2,
"amount": 200.00,
"note": "Bank Transfer Installment"
}'-
Personal Net Savings:
$\text{Total Income} - \text{Total Expenses}$ -
Business Gross Profit:
$\text{Sales Revenue} - \text{Cost of Goods Sold (COGS)}$ -
Conservative Net Worth:
$\text{Unsold Stock Cost Basis} + \text{Cash Collected}$ -
Optimistic Net Worth:
$\text{Fixed Assets} + \text{Unsold Stock Market Value} + \text{Outstanding Receivables} + \text{Cash Collected}$ -
Combined Liquid Cash:
$\text{Personal Net Savings} + \text{Business Cash Collected}$
Run the test suite with uv:
uv run pytest