MCP Hub
Back to servers

mund

AI security scanner - secrets, PII, prompt injection, and exfiltration detection.

Registry
Updated
Feb 18, 2026

Quick Install

npx -y @weave_protocol/mund

🛡️ Weave Protocol Security Suite

npm version license downloads

Enterprise-grade security and orchestration infrastructure for AI agents.

Weave Protocol provides defense-in-depth for autonomous AI systems: secret scanning, secure containment, intent verification, execution replay, multi-agent coordination, and compliance tracking—all with blockchain anchoring for immutable audit trails.

📦 Packages

PackageDescriptionInstall
@weave_protocol/mundGuardian Protocol - Secret & threat scanningnpm i @weave_protocol/mund
@weave_protocol/hordVault Protocol - Secure containment & sandboxingnpm i @weave_protocol/hord
@weave_protocol/domereJudge Protocol - Verification, orchestration & compliancenpm i @weave_protocol/domere
@weave_protocol/witanCouncil Protocol - Consensus, communication & governancenpm i @weave_protocol/witan
@weave_protocol/apiUniversal REST API for all protocolsnpm i @weave_protocol/api

🔌 MCP Registry

Mund is available on the official MCP Registry for Claude Desktop and other MCP-compatible AI assistants.

{
  "mcpServers": {
    "mund": {
      "command": "npx",
      "args": ["@weave_protocol/mund"]
    }
  }
}

View on MCP Registry →


🏗️ Architecture

┌─────────────────────────────────────────────────────────────────────────────┐
│                           WEAVE PROTOCOL SUITE                              │
├─────────────────────────────────────────────────────────────────────────────┤
│                                                                             │
│  ┌───────────────┐  ┌───────────────┐  ┌───────────────┐  ┌─────────────┐   │
│  │     MUND      │  │     HORD      │  │    DŌMERE     │  │    WITAN    │   │
│  │   Guardian    │  │     Vault     │  │     Judge     │  │   Council   │   │
│  ├───────────────┤  ├───────────────┤  ├───────────────┤  ├─────────────┤   │
│  │ • Secrets     │  │ • Storage     │  │ • Intent      │  │ • Consensus │   │
│  │ • PII         │  │ • Redaction   │  │ • Replay      │  │ • Comms Bus │   │
│  │ • Injection   │  │ • Sandbox     │  │ • Handoff     │  │ • Policy    │   │
│  │ • Exfil       │  │ • Encrypt     │  │ • Compliance  │  │ • Recovery  │   │
│  │ • MCP Server  │  │ • Yoxallismus │  │ • Scheduler   │  │ • Voting    │   │ 
│  │               │  │               │  │ • Registry    │  │ • Channels  │   │
│  └───────────────┘  └───────────────┘  └───────────────┘  └─────────────┘   │
│         │                  │                   │                 │          │
│         └──────────────────┴───────────────────┴─────────────────┘          │
│                                   │                                         │
│                    ┌──────────────▼──────────────┐                          │
│                    │     WITAN COUNCIL           │                          │
│                    │  (Orchestrator + N Agents)  │                          │
│                    └──────────────┬──────────────┘                          │
│            ┌────┬────┬────┬────┬──┴──┬────┬────┬────┐                       │
│            ▼    ▼    ▼    ▼    ▼     ▼    ▼    ▼    ▼                       │
│          [A1] [A2] [A3] [A4] [A5]  [A6] [A7] [A8] [...]                     │
│                                                                             │
├─────────────────────────────────────────────────────────────────────────────┤
│                           BLOCKCHAIN LAYER                                  │
│                    ┌─────────────┬─────────────┐                            │
│                    │   Solana    │  Ethereum   │                            │
│                    │   Mainnet   │   Mainnet   │                            │
│                    └─────────────┴─────────────┘                            │
└─────────────────────────────────────────────────────────────────────────────┘

🚀 Quick Start

Option 1: Witan Council (Full Stack)

import { WitanCouncil } from '@weave_protocol/witan';

const council = new WitanCouncil({
  signing_key: 'your-secret-key',
  max_agents: 10
});

await council.start();

// Register agents with voting weights
await council.registerAgent({
  name: 'researcher',
  capabilities: ['search', 'analysis'],
  voting_weight: 2
});

// Submit tasks, propose decisions, send messages
await council.submitTask({ intent: 'Analyze market data', priority: 'high' });

const proposal = await council.propose({
  title: 'Increase compute budget',
  type: 'resource',
  proposer_id: 'researcher'
});

await council.vote(proposal.id, 'researcher', 'approve');

Option 2: Dōmere Orchestration (Core)

import { Orchestrator } from '@weave_protocol/domere';

// Create orchestrator for 10 agents
const orch = new Orchestrator({ max_agents: 10 });
await orch.start();

// Register agents with capabilities
for (let i = 0; i < 10; i++) {
  await orch.registerAgent({
    name: `agent-${i}`,
    capabilities: ['research', 'analysis', 'coding'][i % 3],
    max_concurrent_tasks: 3
  });
}

// Submit tasks with dependencies
const fetchTask = await orch.submitTask({
  intent: 'Fetch Q3 financial data',
  priority: 'high',
  required_capabilities: ['research']
});

const analyzeTask = await orch.submitTask({
  intent: 'Analyze Q3 trends',
  dependencies: [fetchTask.id],  // Waits for fetch to complete
  required_capabilities: ['analysis']
});

// Agents receive tasks via heartbeat
const { tasks_to_run } = await orch.heartbeat('agent-0', []);

Option 3: REST API (Any AI Agent)

npm install @weave_protocol/api
npx weave-api
# Server running on http://localhost:3000
# Scan for secrets/threats
curl -X POST http://localhost:3000/api/v1/mund/scan \
  -H "Content-Type: application/json" \
  -d '{"content": "AWS key: AKIAIOSFODNN7EXAMPLE"}'

# Create verified thread
curl -X POST http://localhost:3000/api/v1/domere/threads \
  -d '{"origin_type": "agent", "origin_identity": "gpt-4", "intent": "Process data"}'

Option 4: Direct Package Usage

import { MundScanner } from '@weave_protocol/mund';
import { HordVault } from '@weave_protocol/hord';
import { ExecutionReplayManager, ComplianceManager } from '@weave_protocol/domere';

// Scan for secrets
const scanner = new MundScanner();
const threats = await scanner.scan('API key: sk-1234567890abcdef');

// Secure storage
const vault = new HordVault();
await vault.store('api-key', 'sk-1234...', { encryption: true });

// Track execution
const replay = new ExecutionReplayManager('encryption-key');
await replay.recordAction({ thread_id: 'thr_1', agent_id: 'agent-1', ... });

🔐 Mund - Guardian Protocol

Real-time threat detection for AI inputs/outputs.

Detects:

  • 🔑 Secrets: API keys (OpenAI, AWS, GitHub, etc.), passwords, tokens
  • 👤 PII: SSN, emails, phone numbers, credit cards
  • 💉 Injection: Prompt injection, jailbreak attempts
  • 📤 Exfiltration: Data theft patterns
import { MundScanner } from '@weave_protocol/mund';

const scanner = new MundScanner();
const result = await scanner.scan(`
  My AWS key is AKIAIOSFODNN7EXAMPLE
  and my SSN is 123-45-6789
`);

console.log(result.issues);
// [
//   { type: 'secret', name: 'AWS Access Key', severity: 'critical' },
//   { type: 'pii', name: 'SSN', severity: 'high' }
// ]

🏛️ Hord - Vault Protocol

Secure containment and sandboxed execution.

Features:

  • 🔒 Encrypted secret storage
  • 📝 Automatic redaction
  • 🏖️ Sandboxed code execution
  • 🔐 Access control policies
import { HordVault } from '@weave_protocol/hord';

const vault = new HordVault({ encryption_key: process.env.VAULT_KEY });

// Store secrets securely
await vault.store('openai-key', 'sk-...', { ttl: 3600 });

// Redact sensitive data
const safe = await vault.redact('My SSN is 123-45-6789');
// → "My SSN is [REDACTED]"

⚖️ Dōmere - Judge Protocol

Intent verification, orchestration, compliance, and blockchain anchoring.

🎯 Intent Tracking & Drift Detection

import { ThreadManager } from '@weave_protocol/domere';

const manager = new ThreadManager();

const thread = await manager.createThread({
  origin_type: 'human',
  origin_identity: 'user@company.com',
  intent: 'Generate quarterly report',
  constraints: ['read-only', 'no-external-api']
});

// Check for drift
const drift = await manager.checkDrift(thread.id, 'Sending data to external API');
// → { drifted: true, reason: 'Violates no-external-api constraint' }

🔄 Execution Replay & Audit Trail

Complete forensic trail with cryptographic verification.

import { ExecutionReplayManager } from '@weave_protocol/domere';

const replay = new ExecutionReplayManager('encryption-key');

// Record every action
await replay.recordAction({
  thread_id: 'thr_xxx',
  agent_id: 'gpt-4-agent',
  agent_type: 'llm',
  action_type: 'inference',
  action_name: 'generate_report',
  input: { prompt: '...' },
  output: { response: '...' },
  latency_ms: 1250,
  cost_usd: 0.03,
  tokens_in: 500,
  tokens_out: 1000
});

// Get tamper-proof trail
const trail = await replay.getExecutionTrail('thr_xxx');
console.log(trail.integrity_valid);  // true
console.log(trail.merkle_root);      // For blockchain anchoring

🤝 Multi-Agent Handoff Verification

Secure delegation between AI agents with permission inheritance.

import { HandoffManager } from '@weave_protocol/domere';

const handoff = new HandoffManager('signing-key', {
  max_delegation_depth: 5,
  max_handoff_duration_ms: 3600000
});

// Delegate from orchestrator to researcher
const token = await handoff.createHandoff({
  thread_id: 'thr_xxx',
  from_agent: 'orchestrator',
  to_agent: 'researcher',
  delegated_intent: 'Find Q3 revenue data',
  constraints: ['read-only', 'internal-data-only'],
  permissions: [{ resource: 'database', actions: ['read'] }],
  max_actions: 10,
  expires_in_ms: 300000
});

// Researcher verifies before acting
const verification = await handoff.verifyHandoff(token.token, 'researcher');

📋 Compliance Checkpoints (SOC2/HIPAA)

Automated compliance tracking and reporting.

import { ComplianceManager } from '@weave_protocol/domere';

const compliance = new ComplianceManager('signing-key');

// HIPAA: Log PHI access
await compliance.logPHIAccess({
  thread_id: 'thr_xxx',
  agent_id: 'medical-assistant',
  patient_id: 'patient_123',
  access_reason: 'Treatment recommendation',
  data_accessed: ['diagnosis', 'medications'],
  legal_basis: 'treatment'
});

// Generate compliance report
const report = await compliance.generateReport({
  framework: 'HIPAA',
  period_start: new Date('2026-01-01'),
  period_end: new Date('2026-03-31')
});
console.log('Score:', report.compliance_score);

📊 Task Scheduler (Multi-Agent)

Priority queue with dependencies, retries, and load balancing.

import { TaskScheduler } from '@weave_protocol/domere';

const scheduler = new TaskScheduler();

const task = await scheduler.createTask({
  intent: 'Analyze Q3 data',
  priority: 'high',
  dependencies: ['fetch-data-task'],
  constraints: {
    required_capabilities: ['data-analysis'],
    max_duration_ms: 300000
  },
  retry_policy: { max_retries: 3, backoff: 'exponential' }
});

const assignment = await scheduler.assignTask(task.id);

🤖 Agent Registry (Health & Capabilities)

Agent lifecycle, heartbeat monitoring, and failover.

import { AgentRegistry } from '@weave_protocol/domere';

const registry = new AgentRegistry();

const agent = await registry.register({
  agent_id: 'agent-7',
  capabilities: ['code-generation', 'testing'],
  max_concurrent_tasks: 3
});

registry.onAgentDown((agent, activeTasks) => {
  console.log(`Agent ${agent.id} down, reassigning tasks`);
});

🗃️ State Manager (Shared State with Locks)

Distributed state with locking, branching, and conflict resolution.

import { StateManager } from '@weave_protocol/domere';

const state = new StateManager({ conflict_resolution: 'last-write-wins' });

// Lock before writing
const lock = await state.acquireLock({ key: 'db', holder: 'agent-3' });
if (lock.acquired) {
  await state.set('db', { updated: true });
  await state.releaseLock('db', 'agent-3');
}

// Git-style branching
await state.createBranch('experiment');
await state.set('config', newConfig, { branch: 'experiment' });
await state.merge('experiment', 'main');

🎛️ Unified Orchestrator

Single interface for multi-agent coordination.

import { Orchestrator } from '@weave_protocol/domere';

const orch = new Orchestrator({ max_agents: 10 });
await orch.start();

for (let i = 0; i < 10; i++) {
  await orch.registerAgent({ name: `worker-${i}`, capabilities: ['general'] });
}

await orch.submitTask({ intent: 'Process batch', priority: 'high' });

const stats = orch.getStats();
console.log(`${stats.agents.ready}/${stats.agents.total} agents ready`);

🏛️ Witan - Council Protocol

Multi-agent consensus, communication, governance, and recovery.

🗳️ Consensus Engine

import { ConsensusEngine } from '@weave_protocol/witan';

const consensus = new ConsensusEngine('signing-key', {
  default_quorum: 0.5,
  default_threshold: 0.6
});

const proposal = await consensus.createProposal({
  title: 'Deploy new model',
  proposal_type: 'action',
  proposer_id: 'orchestrator',
  eligible_voters: ['agent-1', 'agent-2', 'agent-3']
});

await consensus.vote(proposal.id, 'agent-1', 'approve');
await consensus.vote(proposal.id, 'agent-2', 'approve');

const result = await consensus.finalizeProposal(proposal.id);
console.log(result.decision); // 'approved'

📨 Communication Bus

import { CommunicationBus } from '@weave_protocol/witan';

const bus = new CommunicationBus('signing-key');

// Direct message
await bus.send({
  from: 'agent-1',
  to: 'agent-2',
  type: 'data-handoff',
  payload: { dataset_id: 'ds_123' }
});

// Broadcast to all
await bus.broadcast({
  from: 'orchestrator',
  type: 'priority-change',
  payload: { all_tasks: 'high' }
});

📜 Policy Engine

import { PolicyEngine } from '@weave_protocol/witan';

const policy = new PolicyEngine();

// Rate limit: 100 requests per minute
await policy.createRateLimit({
  name: 'api-limit',
  targets: [{ type: 'all' }],
  max_requests: 100,
  window_ms: 60000
});

// Enforce
const decision = await policy.enforce({
  agent_id: 'agent-1',
  action: 'api_call',
  timestamp: new Date()
});

🔄 Recovery Manager

import { RecoveryManager } from '@weave_protocol/witan';

const recovery = new RecoveryManager('signing-key');

// Checkpoint
const checkpoint = await recovery.checkpoint({
  name: 'Pre-deployment',
  created_by: 'admin'
});

// Transaction with auto-rollback
const txn = await recovery.beginTransaction({
  initiator: 'agent-1',
  auto_checkpoint: true
});

// ... operations ...
await recovery.commitTransaction(txn.id);
// or: await recovery.rollbackTransaction(txn.id);

⛓️ Blockchain Deployments

ChainNetworkContract/ProgramExplorer
SolanaMainnet6g7raTAHU2h331VKtfVtkS5pmuvR8vMYwjGsZF1CUj2oView
SolanaDevnetBeCYVJYfbUu3k2TPGmh9VoGWeJwzm2hg2NdtnvbdBNCjView
EthereumMainnet0xAA8b52adD3CEce6269d14C6335a79df451543820View

📊 Feature Matrix

FeatureMundHordDōmereWitan
Secret Detection
PII Detection
Injection Detection
MCP Server
Encrypted Storage
Redaction
Sandboxing
Yoxallismus Cipher
Intent Tracking
Drift Detection
Execution Replay
Multi-Agent Handoff
SOC2 Compliance
HIPAA Compliance
PCI-DSS Compliance
ISO27001 Compliance
Task Scheduling
Agent Registry
Shared State/Locks
Blockchain Anchoring
Consensus/Voting
Agent Messaging
Policy Engine
Checkpoints/Recovery

🗺️ Roadmap

Current (v1.x)

  • ✅ Mund - Secret & threat scanning
  • ✅ Hord - Secure vault & sandbox
  • ✅ Dōmere - Verification & orchestration
  • ✅ Witan - Consensus, communication & governance
  • ✅ REST API
  • ✅ Ethereum mainnet deployment
  • ✅ Solana mainnet deployment
  • ✅ PCI-DSS & ISO27001 compliance frameworks
  • ✅ Yoxallismus Vault Cipher (Hord)
  • ✅ MCP server integration (Mund)

Next (v2.x)

  • 🤖 Advanced agent coordination patterns - In progress
  • 📊 Real-time monitoring dashboard - In research

📄 License

Apache 2.0 - See LICENSE for details.


🤝 Contributing

Contributions welcome! Here's how:

  1. Fork the repository
  2. Create a feature branch (git checkout -b feature/amazing-feature)
  3. Commit your changes (git commit -m 'Add amazing feature')
  4. Push to the branch (git push origin feature/amazing-feature)
  5. Open a Pull Request

Please ensure your code passes existing tests and follows the project's coding style.


Made with ❤️ for AI Safety

Reviews

No reviews yet

Sign in to write a review