# VITNA – Containment for rogue AI agents Intercepts stdio MCP server calls and halts risky tool actions until human approval. Keeps AI agents inside explicit execution boundaries. - MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI - Author: COSTRINITY - License: MIT · policy=link-only · description only; the file stays with the author - Source: https://github.com/COSTRINITY/vitna-compliance-mcp - Updated: 2026-10-14 - HTML: https://affpapa.org/en/nevibecode/mcp/vitna-containment-for-rogue-ai-agents ## What it does VITNA wraps stdio MCP servers to inspect tool calls before execution. When a client triggers `tools/call`, the guard checks the requested tool against a JSON policy file set in `VITNA_GUARD_POLICY`. If an action matches hold rules, VITNA pauses the process until an operator confirms it in the web dashboard or the holding window expires. Beyond stdio wrapping, the package provides 23 preflight compliance tools covering 24 statutory frameworks across 13 legal jurisdictions. Each check yields an immediate status of `allowed`, `blocked`, or `flagged`. Claiming an account with a passkey or GitHub enables cryptographically signed records that can be verified offline. The guard only controls servers wrapped behind its process. A client's native terminal shell or any server launched without the guard bypasses enforcement completely. ## Requirements - Needs: Node.js (commands run via npx) - Remote server: `https://vitna.costrinity.xyz/api/mcp` - Environment variables: `VITNA_OWNER_ID`, `VITNA_API_KEY`, `VITNA_EMAIL`, `VITNA_BASE_URL`, `VITNA_OPEN_CLAIM` — keys and tokens: ask the human for the values, never commit them or put them in shared files. ## Install ### Claude Code (native platform) 1. **Platform CLI: remote server** ```bash claude mcp add --transport http vitna-compliance-preflight https://vitna.costrinity.xyz/api/mcp ``` 2. **Platform CLI** ```bash claude mcp add --env VITNA_OWNER_ID= --env VITNA_API_KEY= --env VITNA_EMAIL= --env VITNA_BASE_URL= --env VITNA_OPEN_CLAIM= --transport stdio vitna-containment-for-rogue-ai-agents -- npx -y @costrinity/vitna-compliance-mcp ``` Replace <…> with your own values ### Codex 1. **Platform CLI: remote server** ```bash codex mcp add vitna-containment-for-rogue-ai-agents --url https://vitna.costrinity.xyz/api/mcp ``` 2. **Platform CLI** ```bash codex mcp add vitna-containment-for-rogue-ai-agents --env VITNA_OWNER_ID= --env VITNA_API_KEY= --env VITNA_EMAIL= --env VITNA_BASE_URL= --env VITNA_OPEN_CLAIM= -- npx -y @costrinity/vitna-compliance-mcp ``` Replace <…> with your own values ### Cursor **Config file** File `.cursor/mcp.json`: ```json { "mcpServers": { "vitna-containment-for-rogue-ai-agents": { "url": "https://vitna.costrinity.xyz/api/mcp" } } } ``` File `.cursor/mcp.json`: ```json { "mcpServers": { "vitna-containment-for-rogue-ai-agents": { "command": "npx", "args": [ "-y", "@costrinity/vitna-compliance-mcp" ], "env": { "VITNA_OWNER_ID": "", "VITNA_API_KEY": "", "VITNA_EMAIL": "", "VITNA_BASE_URL": "", "VITNA_OPEN_CLAIM": "" } } } } ``` Replace <…> with your own values ### GitHub Copilot **Config file** File `.vscode/mcp.json`: ```json { "servers": { "vitna-containment-for-rogue-ai-agents": { "type": "http", "url": "https://vitna.costrinity.xyz/api/mcp" } } } ``` File `.vscode/mcp.json`: ```json { "servers": { "vitna-containment-for-rogue-ai-agents": { "type": "stdio", "command": "npx", "args": [ "-y", "@costrinity/vitna-compliance-mcp" ], "env": { "VITNA_OWNER_ID": "", "VITNA_API_KEY": "", "VITNA_EMAIL": "", "VITNA_BASE_URL": "", "VITNA_OPEN_CLAIM": "" } } } } ``` Replace <…> with your own values ### Gemini CLI 1. **Platform CLI: remote server** ```bash gemini mcp add -t http vitna-containment-for-rogue-ai-agents https://vitna.costrinity.xyz/api/mcp ``` 2. **Platform CLI** ```bash gemini mcp add -e VITNA_OWNER_ID= -e VITNA_API_KEY= -e VITNA_EMAIL= -e VITNA_BASE_URL= -e VITNA_OPEN_CLAIM= vitna-containment-for-rogue-ai-agents npx -y @costrinity/vitna-compliance-mcp ``` Replace <…> with your own values ## When to use it - pause risky file modifications until a human approves them - evaluate agent actions against statutory compliance rules - keep signed audit receipts of tool execution decisions - block unapproved tool calls across stdio MCP servers ## Example prompts - Wrap the filesystem server using VITNA guard and `policy.json` - Run a preflight compliance check before editing this configuration ## Similar - [rail-score](https://affpapa.org/en/nevibecode/mcp/rail-score.md) — MCP server - [speclock](https://affpapa.org/en/nevibecode/mcp/speclock.md) — MCP server · ★ 25 - [Aegis — AI Agent Governance](https://affpapa.org/en/nevibecode/mcp/aegis-ai-agent-governance.md) — MCP server · ★ 18 - [Aletheia Runtime Safety & Scope Creep Filter](https://affpapa.org/en/nevibecode/mcp/aletheia-runtime-safety-scope-creep-filter.md) — MCP server · ★ 12 - [Aperion Shield](https://affpapa.org/en/nevibecode/mcp/aperion-shield.md) — MCP server · ★ 11 - [nah](https://affpapa.org/en/nevibecode/plugins/nah.md) — plugin · ★ 493 --- NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md