Built an MCP server with 12 decision intelligence tools -- complements Context7 nicely

[Whatsonyourmind]

Hey Context7 community! I built an MCP server that pairs well with Context7 in AI coding workflows.

The idea

Context7 gives your AI agent the right documentation context. OraClaw gives it structured decision-making capabilities -- optimization, prediction, risk analysis, and planning algorithms.

Together, an agent can both understand the docs (Context7) and make mathematically-sound decisions (OraClaw).

Setup alongside Context7

{
  "mcpServers": {
    "context7": {
      "command": "npx",
      "args": ["-y", "@upstash/context7-mcp@latest"]
    },
    "oraclaw": {
      "command": "npx",
      "args": ["@oraclaw/mcp-server"]
    }
  }
}

The 12 MCP tools

Tool	What It Does
optimize_bandit	A/B test selection (UCB1, Thompson, Epsilon-Greedy)
optimize_contextual	Context-aware selection (LinUCB)
optimize_cmaes	Continuous optimization (CMA-ES)
solve_constraints	LP/MIP optimization via HiGHS
solve_schedule	Energy-matched task scheduling
analyze_decision_graph	PageRank and community detection
analyze_portfolio_risk	VaR and CVaR (Expected Shortfall)
score_convergence	Multi-source agreement scoring
score_calibration	Prediction quality (Brier/log score)
predict_forecast	ARIMA and Holt-Winters forecasting
detect_anomaly	Z-Score and IQR anomaly detection
plan_pathfind	A* pathfinding

Every tool is deterministic, sub-25ms, returns structured JSON. No API keys needed.

MIT licensed, 1,072 tests, 15 npm packages.

Links: GitHub | npm | Demo