Deterministic Agent Control
Governance gateway for AI agents — bounded, auditable, session-aware control with MCP proxy, shell proxy & HTTP API. Works with Cursor, Claude Code, Codex, and any MCP-compatible agent.
What is Deterministic Agent Control?
Deterministic Agent Control is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to governance gateway for ai agents — bounded, auditable, session-aware control with mcp proxy, shell proxy & http api. works with cursor, claude code, codex, and any mcp-compatible agent.
Governance gateway for AI agents — bounded, auditable, session-aware control with MCP proxy, shell proxy & HTTP API. Works with Cursor, Claude Code, Codex, and any MCP-compatible agent.
This server falls under the Security category on MCPgee, the world's largest MCP server directory with 33,000+ servers.
Features
- Governance gateway for AI agents — bounded, auditable, sessi
Use Cases
Maintainer
Works with
Installation
Manual Installation
npx deterministic-agent-control-protocolConfiguration
Configuration Details
claude_desktop_config.json
Performance
Response Metrics
Resource Usage
How to Set Up and Use Deterministic Agent Control
The Deterministic Agent Control Protocol (det-acp) is a governance gateway that intercepts and evaluates every action an AI agent attempts before it executes. It enforces policy-defined boundaries — file paths, shell commands, cost budgets, and human-approval gates — and records everything in a tamper-evident audit ledger. Teams use it to run Cursor, Claude Code, Codex, and other MCP-compatible agents in production environments where unbounded agent behavior is unacceptable.
Prerequisites
- Node.js 18 or later installed
- An MCP-compatible client such as Claude Desktop, Cursor, or Claude Code
- A policy file (agent.policy.yaml) defining allowed capabilities and limits
- npm or npx available on PATH
Install the det-acp package
Install the core package globally or use npx for zero-install usage.
npm i -g @det-acp/coreInitialize the agent policy for your client
Run the init command for your specific agent client. This generates a starter agent.policy.yaml and wires the MCP proxy into your client's config.
npx det-acp init claude-code
# or for Cursor:
npx det-acp init cursor
# or for Codex:
npx det-acp init codexEdit your agent.policy.yaml
Define which tools the agent may use, set runtime limits, declare gates for actions requiring human approval, and list forbidden patterns such as .env files or destructive shell commands.
# agent.policy.yaml example
capabilities:
- file:read
- file:write
- git:diff
- git:commit
limits:
maxActions: 50
maxCost: 0.50
gates:
- command:run
forbidden:
- ".env"
- "rm -rf"Start the MCP proxy server
Launch det-acp as an MCP proxy that your agent client connects to. All tool calls are evaluated against the policy before being forwarded to the underlying tools.
npx det-acp proxy --policy ./agent.policy.yamlConfigure your MCP client
Add det-acp to your Claude Desktop or Cursor MCP server configuration so the client routes all tool calls through the governance gateway.
Run the HTTP API server (optional)
For programmatic governance from CI pipelines or dashboards, start the HTTP server on port 3100.
npx det-acp serve --port 3100Deterministic Agent Control Examples
Client configuration
Add det-acp as an MCP server in Claude Desktop so all agent tool calls pass through the governance gateway.
{
"mcpServers": {
"det-acp": {
"command": "npx",
"args": ["det-acp", "proxy", "--policy", "./agent.policy.yaml", "--evolve"]
}
}
}Prompts to try
Example prompts that demonstrate governance-aware agent interactions.
- "Refactor the authentication module but do not touch any .env files"
- "Run the test suite and commit the passing result with a descriptive message"
- "List all files changed in the last git commit and explain the diff"
- "Attempt to delete the build artifacts directory and request approval if needed"Troubleshooting Deterministic Agent Control
Agent action is denied even though it seems allowed in the policy
Check the audit ledger in ./ledgers/ for the exact verdict and the matched forbidden pattern. The policy evaluator uses glob matching, so a pattern like '*.env' will block any file ending in .env.
Policy self-evolution suggestions are not appearing
Ensure you started the proxy with the --evolve flag. Without it, denied actions are logged but no relaxation suggestions are generated.
npx det-acp init does not generate a config file
Run the command from the project root directory where you want the policy file placed. Check that Node.js 18+ is active with 'node --version'.
Frequently Asked Questions about Deterministic Agent Control
What is Deterministic Agent Control?
Deterministic Agent Control is a Model Context Protocol (MCP) server that governance gateway for ai agents — bounded, auditable, session-aware control with mcp proxy, shell proxy & http api. works with cursor, claude code, codex, and any mcp-compatible agent. It connects AI assistants to external tools and data sources through a standardized interface.
How do I install Deterministic Agent Control?
Follow the installation instructions on the Deterministic Agent Control GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.
Which AI clients work with Deterministic Agent Control?
Deterministic Agent Control works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.
Is Deterministic Agent Control free to use?
Yes, Deterministic Agent Control is open source and available under the MIT license. You can use it freely in both personal and commercial projects.
Deterministic Agent Control Alternatives — Similar Security Servers
Looking for alternatives to Deterministic Agent Control? Here are other popular security servers you can use with Claude, Cursor, and VS Code.
Casdoor
★ 13.6kAn open-source Agent-first Identity and Access Management (IAM) /LLM MCP & agent gateway and auth server with web UI supporting OpenClaw, MCP, OAuth, OIDC, SAML, CAS, LDAP, SCIM, WebAuthn, TOTP, MFA, Face ID, Google Workspace, Azure AD
ghidraMCP
★ 9.0kAn Model Context Protocol server that enables LLMs to autonomously reverse engineer applications by exposing Ghidra's decompilation and analysis tools. It allows AI agents to list code structures, rename methods, and analyze binaries directly through
HexStrike AI
★ 8.9kHexStrike AI MCP Agents is an advanced MCP server that lets AI agents (Claude, GPT, Copilot, etc.) autonomously run 150+ cybersecurity tools for automated pentesting, vulnerability discovery, bug bounty automation, and security research. Seamlessly b
IDA Pro MCP
★ 8.7kEnables AI-assisted reverse engineering in IDA Pro by providing tools to analyze binaries, decompile functions, manage comments, search patterns, and interact with the IDA database through natural language.
Anthropic Cybersecurity Skills
★ 6.6k754 structured cybersecurity skills for AI agents · Mapped to 5 frameworks: MITRE ATT&CK, NIST CSF 2.0, MITRE ATLAS, D3FEND & NIST AI RMF · agentskills.io standard · Works with Claude Code, GitHub Copilot, Codex CLI, Cursor, Gemini CLI & 20+ platform
Hooker
★ 5.1k🔥🔥 hooker is a Frida-based reverse engineering toolkit for Android. It offers a user-friendly CLI, universal scripts, auto hook generation, memory roaming to detect activities/services, one-click SOCKS5 proxy setup, Frida JustTrustMe, and BoringSSL u
Browse More Security MCP Servers
Explore all security servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.
Set Up Deterministic Agent Control in Your Editor
Choose your AI client for step-by-step setup instructions.
Quick Config Preview
Add this to your claude_desktop_config.json or .cursor/mcp.json
Ready to use Deterministic Agent Control?
Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.