Autonomous Trading Bot with Storacha MCP – June 4, 2025 – PLDG Cohort-3 #122

adielliot37 · 2025-06-03T20:34:42Z

adielliot37
Jun 3, 2025
Collaborator

Integrated Storacha MCP REST server and verified Base64-based upload/retrieve via tools/call → upload and retrieve.
Implemented delegation management placeholders for upcoming UCAN-based authorization to secure uploads.
Fetched top-20 USDT pairs from Binance, computed multi-indicator signals (RSI, MACD, EMA, Bollinger, Stochastic, ADX, CCI) via pandas_ta.
Executed \$10 USD, 5× leverage market orders on Binance Futures (USDT-M) when Long/Short signals were decisive.
Generated concise trade reports and uploaded them to Storacha, storing `{symbol, datetime, CID}` in trades.json.

Completed end-to-end multi-agent flow:
- Agent A: computes technical signal and Token Metrics AI overview.
- Agent B: opens a Binance Futures trade when signal is non-neutral.
- Agent C: formats and uploads the concise trade report to Storacha.
What went well:
- MCP REST uploads returned valid CIDs and retrieval works reliably.
- Indicator-based signals produced consistent Long/Short decisions tested on multiple pairs.
- Telegram bot interface responds to /auto and /trades commands as expected.
What was challenging:
- Ensuring pandas_ta compatibility with NumPy (patched np.NaN → np.nan).
- Handling Binance Futures API edge cases (zero-price pairs, leverage errors).
- Formatting concise reports while preserving critical trade parameters and AI summary.

How can we streamline UCAN delegation issuance and rotation across multiple agents to minimize friction?
Is there a way to register custom MCP tools or aliases in Storacha for easier agent usage?

Integrate UCAN-based authorization to enforce permissioned MCP uploads per agent.
Add additional indicators (e.g., VWAP, Supertrend) and implement weighted signal tuning.
Implement retry logic and exponential backoff for MCP REST calls and Binance API requests.
Enhance /trades view to filter by symbol or date range and add search capability.