What can you do with this server?

The ORACLE-H MCP server provides a human-in-the-loop validation gateway, requiring human approval via Telegram before AI agents execute critical, destructive, or irreversible actions. Core Tools: * oracle_validate — Submit an action for human approval, providing: * A precise technical description, rationale, and projected impact * An optional financial amount at stake (USD) * A configurable TTL/timeout (default: 5 minutes) * Returns a request_id for tracking * oracle_poll_status — Check the status of a submitted request (poll every 10–15s), receiving one of: * approved → proceed * rejected → abort * expired → no response in time; abort or retry Key Use Cases: * Destructive operations (deleting files, dropping databases, force-pushing to git) * External writes (sending emails, posting to APIs) * Financial transactions (payments, billing changes) * Infrastructure changes (deployments, DNS, environment variables) Additional Highlights: * Increases AI workflow reliability from ~81.5% to ~99.6% by adding human checkpoints * Secured via bearer token authentication and HMAC-SHA256 signed webhooks * Compatible with Claude Desktop, Cursor, VS Code extensions (Cline, Roo Code, GitHub Copilot), n8n, Zed, Continue.dev, and LangGraph

Which integrations are available for this server?

Enables human-on-the-loop validation by sending notifications to Telegram, allowing users to approve or reject critical, destructive, or irreversible actions before an AI agent executes them.

mcp-oracle-h

by tponscr-debug

Overview Schema Source Related Servers Discussions Admin

JavaScript

Remote

ORACLE-H MCP Server

Human-on-the-Loop validation for autonomous AI agents — via Model Context Protocol

What is ORACLE-H?

ORACLE-H is a safety infrastructure that acts as a mandatory checkpoint for autonomous AI agents. Before executing any critical, destructive, or irreversible action, the agent submits it for human approval via Telegram. The human decides with a single tap.

Agent → oracle_validate → Human gets Telegram alert → Approve/Reject → Agent proceeds

Without ORACLE-H: 4-step workflow succeeds ~81.5% of the time. With ORACLE-H: Same workflow succeeds ~99.6% of the time.

MCP Tools

Tool	Description
`oracle_validate`	Submit a risky action for human approval. Returns a `request_id`.
`oracle_poll_status`	Poll the decision until approved, rejected, or expired.

# Test immediately with the public demo key
ORACLE_SIGNATURE_KEY=demo-signature-key-change-in-production
ORACLE_BASE_URL=https://oracle-h.online

Installation

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

{
  "mcpServers": {
    "oracle-h": {
      "command": "npx",
      "args": ["-y", "oracle-h-mcp"],
      "env": {
        "ORACLE_SIGNATURE_KEY": "<your-signature-key>",
        "ORACLE_BASE_URL": "https://oracle-h.online"
      }
    }
  }
}

Cursor / Windsurf

Add to your MCP settings:

{
  "oracle-h": {
    "command": "npx",
    "args": ["-y", "oracle-h-mcp"],
    "env": {
      "ORACLE_SIGNATURE_KEY": "<your-signature-key>",
      "ORACLE_BASE_URL": "https://oracle-h.online"
    }
  }
}

Cline (VS Code)

// File: cline_mcp_settings.json
{
  "mcpServers": {
    "oracle-h": {
      "command": "npx",
      "args": ["-y", "oracle-h-mcp"],
      "env": {
        "ORACLE_SIGNATURE_KEY": "demo-signature-key-change-in-production",
        "ORACLE_BASE_URL": "https://oracle-h.online"
      }
    }
  }
}

Roo Code (VS Code)

// File: .roo/mcp.json (project-level)
{
  "mcpServers": {
    "oracle-h": {
      "command": "npx",
      "args": ["-y", "oracle-h-mcp"],
      "env": {
        "ORACLE_SIGNATURE_KEY": "demo-signature-key-change-in-production",
        "ORACLE_BASE_URL": "https://oracle-h.online"
      }
    }
  }
}

GitHub Copilot (VS Code)

Add to VS Code settings.json:

{
  "github.copilot.chat.mcp.servers": {
    "oracle-h": {
      "command": "npx",
      "args": ["-y", "oracle-h-mcp"],
      "env": {
        "ORACLE_SIGNATURE_KEY": "demo-signature-key-change-in-production",
        "ORACLE_BASE_URL": "https://oracle-h.online"
      }
    }
  }
}

n8n

Use the MCP Client Tool node in n8n, running via supergateway:

npx -y supergateway --stdio "npx -y oracle-h-mcp" --port 8811

Then connect n8n's MCP Client Tool to http://localhost:8811/sse

Zed Editor

Add to settings.json:

{
  "context_servers": {
    "oracle-h": {
      "command": {
        "path": "npx",
        "args": ["-y", "oracle-h-mcp"]
      },
      "env": {
        "ORACLE_SIGNATURE_KEY": "<your-signature-key>",
        "ORACLE_BASE_URL": "https://oracle-h.online"
      }
    }
  }
}

Continue.dev

Create .continue/mcpServers/oracle-h.yaml:

name: oracle-h
version: 1.0.0
schema: v1
mcpServers:
  - name: oracle-h
    command: npx
    args:
      - -y
      - oracle-h-mcp
    env:
      ORACLE_SIGNATURE_KEY: <your-signature-key>
      ORACLE_BASE_URL: https://oracle-h.online

LangGraph (Python)

from langchain_mcp_adapters.client import MultiServerMCPClient

async with MultiServerMCPClient({
    "oracle-h": {
        "command": "npx",
        "args": ["-y", "oracle-h-mcp"],
        "env": {
            "ORACLE_SIGNATURE_KEY": "your-key",
            "ORACLE_BASE_URL": "https://oracle-h.online"
        },
        "transport": "stdio"
    }
}) as client:
    tools = client.get_tools()

How it works

Agent calls oracle_validate with action, rationale, projected impact, and TTL
Human receives a Telegram notification with full context and risk analysis
Human taps Approve or Reject directly in Telegram
Agent polls oracle_poll_status until decision arrives
Agent proceeds or aborts based on the human decision

The system includes:

🛡️ Anti-hallucination Watchdog — detects context truncation before submitting
📊 Reliability Calculator — shows the statistical improvement (81.5% → 99.6%)
⏳ TTL / Expiry — requests auto-expire if no human response within the configured timeout
🔐 Bearer Token Auth — each agent has a unique signature key
🔒 HMAC-SHA256 signed webhooks — tamper-proof decisions

System Prompt Snippet

Add this to your agent's system prompt to ensure ORACLE-H is always used correctly:

## Human Approval Gate — ORACLE-H (MANDATORY)

Call oracle_validate BEFORE executing any action in these categories:

DESTRUCTIVE: Deleting files, dropping databases, removing users, git force-push
EXTERNAL WRITES: Sending emails, posting to APIs, submitting forms
FINANCIAL: Payments, billing changes, subscription modifications
INFRASTRUCTURE: Deployments, server config, DNS, environment variables

POLLING PROTOCOL:
After oracle_validate, poll oracle_poll_status every 10-15 seconds until:
- "approved" → proceed
- "rejected" → STOP, inform user, propose alternatives
- "expired"  → STOP, ask user if they want to retry

HARD RULES:
- NEVER execute before "approved"
- If oracle_validate is unreachable: BLOCK the action, inform user
- Read-only operations do NOT require oracle_validate

API

Full API documentation: oracle-h.online/llms.txt OpenAPI spec: oracle-h.online/openapi.json

License

MIT

Available Tools

2 tools

oracle_poll_statusA

Check the current status of a validation request previously submitted via oracle_validate. Poll every 10-15 seconds until status is "approved" or "rejected". If "expired", the human did not respond in time — abort the action.

Inputs: request_id (required).

oracle_validateA

Submit a critical or irreversible action to ORACLE-H for human approval. The human operator will receive a Telegram notification with full context and can approve or reject. ALWAYS use this tool before executing any destructive, financial, or irreversible operation. Returns a request_id to use with oracle_poll_status to retrieve the decision.

Inputs: action (required), rationale (required), projected_impact (required), financial_amount, +1 more.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

Deploy Server

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

A4.2/5.0

Scored 2026-04-17 19:21 across 2 tools

Disambiguation5/5

The two tools have completely distinct and non-overlapping purposes: oracle_validate is for submitting actions for human approval, while oracle_poll_status is for checking the status of those submissions. Their descriptions clearly differentiate them, with no ambiguity in functionality.

Naming Consistency5/5

Both tools follow a consistent 'oracle_verb_noun' pattern (oracle_poll_status and oracle_validate), using snake_case and starting with the server prefix. This makes them predictable and easy to identify within the server's namespace.

Tool Count2/5

With only two tools, the server feels thin for its apparent purpose of handling human approval workflows. While the tools cover the core submit-and-poll cycle, there are likely missing operations like listing pending requests or managing notifications, making the scope underdeveloped.

Completeness3/5

The server provides basic coverage for the human approval domain with submit and poll operations, but there are notable gaps. Missing tools for operations like canceling requests, viewing request history, or configuring approval settings limit its completeness, though agents can still work with the core workflow.

Resources

Related MCP Connectors

penholder
ai.penholder
Preventive human-approval write-gate for AI agents: writes commit only after a human approves.
Security Autonomous Agents Agent Orchestration
1
https://api.penholder.ai/mcp
X-EGOA
com.x-ego
Human approval for irreversible AI agent actions, bound to the exact tool call by a passkey tap
Autonomous Agents Security
41
https://mcp.x-ego.com/mcp
psh.nowA
io.github.passy-ai
Human approval for AI agents: one ask, one person, one boolean.
Autonomous Agents Security
11
https://psh.now/mcp
withHumanA
ai.withhuman
Set up human approval for your AI agents' tool calls.
Agent Orchestration Security
8412
https://app.withhuman.ai/api/mcp/v1