Whatsonyourmind/oraclaw

Overview

Oraclaw is a deterministic decision-intelligence MCP server for AI agents. It provides 17 tools and 21 algorithms (including LinUCB, HiGHS LP/MIP, PageRank, Monte Carlo, CMA-ES, and conformal) with sub-25ms response times and zero LLM cost. It has been field-validated in over 12 open-source projects and holds an AAA rating on Glama.

Best for

Best for
Developers building deterministic, low-latency decision agents in TypeScript

Use cases

• Embedding decision-making logic into AI agents without LLM overhead
• Running optimization or ranking algorithms (e.g., PageRank, Monte Carlo) via MCP
• Building deterministic, low-latency agent workflows for production systems

How to use

Install

npm install @oraclaw/bandit

Tools exposed

• optimize_bandit
• optimize_contextual
• optimize_evolve
• solve_schedule
• score_convergence
• score_calibration
• predict_bayesian
• predict_ensemble
• plan_pathfind
• simulate_montecarlo
• simulate_scenario
• optimize_cmaes
• solve_constraints
• analyze_graph
• analyze_risk
• predict_forecast
• detect_anomaly

Example client config

{\n  "mcpServers": {\n    "oraclaw": {\n      "command": "npx",\n      "args": ["-y", "@oraclaw/mcp-server"]\n    }\n  }\n}

Notes

Oraclaw is a deterministic decision-intelligence MCP server for AI agents. It provides 17 tools and 21 algorithms (including LinUCB, HiGHS LP/MIP, PageRank, Monte Carlo, CMA-ES, and conformal) with sub-25ms response times and zero LLM cost. It has been field-validated in over 12 open-source projects and holds an AAA rating on Glama.

8 stars on GitHub. Last updated 2026-06-01. Licensed MIT.

Use cases

• Embedding decision-making logic into AI agents without LLM overhead
• Running optimization or ranking algorithms (e.g., PageRank, Monte Carlo) via MCP
• Building deterministic, low-latency agent workflows for production systems

Pros

• Sub-25ms latency enables real-time agent decisions
• Zero LLM cost reduces operational expenses
• Broad algorithm library (21 algorithms) covers many decision-intelligence needs

Cons

• Limited to deterministic algorithms, not suitable for probabilistic or generative tasks
• Requires MCP-compatible agent infrastructure to integrate
• Small community and ecosystem relative to larger agent frameworks

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.