Fincept Ai Ops

🏗️ Fincept AI Ops

> Research-first, supervised paper trading system. > FastAPI + N8N + MCP. 5 active connectors. Enterprise grade.

Status: 🟢 Production-Ready — Audit Score 8.4/10 Default Branch: main ← canonical, fully up-to-date Owner: Mehmet Zeki (ZeZilly) Version: 1.0 / MVP Last Updated: 2026-05-22

---

⚠️ Branch Status

| Branch | Role | Content | Status | |---|---|---|---| | main | Canonical / Production | Full implementation | ✅ Default & Active | | youtube | Legacy dev branch | Identical to main | 🔒 Frozen (no divergence) |

> main contains 100% of all implementation. youtube was the development branch during initial build — all content was squash-merged into main on 2026-05-22. Use main for all future work.

---

Core Principle

> No live trading. No execution without human approval. No broker action without audit log.

---

System Overview

Market Data + News + Fundamentals
            ↓
      Research Pipeline
            ↓
      Strategy Lab (signal candidate)
            ↓
      Risk Policy (evaluate)
            ↓
    Human Approval Gate
            ↓
    Paper Broker Execution
            ↓
    Audit Logger + State Store
            ↓
    Daily Briefing Generator

---

5 Active Connectors (MVP)

| # | Connector | Purpose | Status | |---|---|---|---| | 1 | marketdata | Price, OHLCV, watchlist | ✅ Active | | 2 | fundamentals | Balance sheet, income, cash flow | ✅ Active | | 3 | news | News and event data | ✅ Active | | 4 | backtest | Signal test and performance | ✅ Active | | 5 | brokersandbox | Paper order execution | ✅ Active |

All other connectors are disabled in MVP.

---

What's Implemented (v1.0)

Agent Layer

• risk_guard — Portfolio-context-enriched risk evaluation
• execution_ops — Dual-gate: risk check + human approval before execution

MCP Server

• mcp/server.py — Full StdIO MCP server with 7 tools
• mcp/tools/market_data.py
• mcp/tools/research_execution.py

API & Middleware

• FastAPI app with rate limiting + request size limits
• CORS env-driven configuration
• Tenacity retry decorator on connectors
• Env-configurable risk limits (risk_policy.py)

Infrastructure

• Dockerfile — Production-ready with healthcheck
• docker-compose.yml — API + MCP services
• .github/workflows/ci.yml — Test matrix (Python 3.11 / 3.12 + Docker build)

Tests (17 cases)

• test_risk_guard_agent.py (4)
• test_execution_ops_agent.py (4)
• test_security_middleware.py (3)
• test_mcp_server.py (6)

---

Quick Start

pip install -r requirements.txt
cp .env.example .env
uvicorn apps.fincept_aiops.app:app --reload --port 8000
pytest -q

---

Startup Sequence

1. Create storage dirs
2. Start FastAPI app (port 8000)
3. Start MCP server (python mcp/server.py)
4. Import N8N workflows
5. Run smoke tests (pytest -q)
6. Activate paper flows

---

Hard Limits

• ❌ No live trading
• ❌ No prompt-based risk override
• ❌ No broker action without audit
• ❌ Max 6 agents
• ❌ Max 5 active connectors
• ❌ Max 1 risk policy source

---

Realization Scope

| Phase | Scope | Decision | |---|---|---| | 1 | Research + briefing + paper broker | ✅ MVP | | 2 | Backtest + screening + approval gate | ✅ MVP+ | | 3 | Sandboxed broker integration | ⏳ Pilot | | 4 | Live trading | ❌ No-Go | | 5 | Multi-broker, full automation | 💤 V2 |

---

Docs

See [docs/](./docs/) for full architecture, API contracts, deployment guide, ops playbook, and decision matrix.

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Zekiog
• Source: Zekiog/fincept-ai-ops
• License: MIT

Install and usage instructions live in the source repository linked above.