Governance infrastructure for autonomous AI systems.
Sentinel SCA is governance infrastructure for AI agents that enforces security policies, records actions in a tamper-evident ledger, and enables forensic replay of autonomous systems.
Sentinel sits between autonomous agents and real infrastructure, verifying identity, evaluating policy decisions, and governing how AI actions are executed.
AI agents are increasingly able to perform real infrastructure actions:
- restarting services
- deploying code
- modifying databases
- executing cloud operations
Without a control boundary, these actions can become unsafe.
Sentinel SCA introduces a governance layer between AI systems and infrastructure, ensuring every action is authenticated, evaluated, and recorded.
Every AI action passes through a deterministic governance pipeline.
AI Agent
↓
Agent Identity (Ed25519)
↓
Capability Tokens
↓
Schema Validation
↓
Behavior Limits
↓
Policy Engine
↓
Deterministic Action Hash
↓
Execution Worker
↓
AI Action Report
↓
Execution Ledger
↓
Hash-Chained Audit Log
↓
Live Action Timeline
↓
Action Replay & Forensics
Agents authenticate using Ed25519 cryptographic signatures, ensuring every action is attributable to a specific autonomous system.
Every request is evaluated against explicit policy rules:
ALLOW
REVIEW
DENY
High-risk actions are automatically routed for human approval.
Approved actions are executed only through Sentinel execution workers.
All actions are recorded in a hash-chained ledger preventing silent modification.
Operators can observe real-time agent activity and decisions.
Any historical event can be reconstructed to understand what happened and why.
Sentinel continuously evaluates agent behavior and assigns security scores based on governance outcomes and risk patterns.
Agent: deploy-bot-7
Identity: ed25519:f83a...
Capability: restart_service
Target: redis
Decision: REVIEW
Policy Version: v2
Timestamp: 2026-03-15T16:45Z
Action Hash: 42655ff1...
Previous Hash: 482248ea...
AI DevOps Agents
Autonomous Infrastructure Systems
Multi-Agent Environments
Security & Compliance Monitoring
git clone https://github.com/sentinelSCA/sentinel.git
cd sentinel
Documentation is available in:
docs/
docs/protocol.md
docs/architecture.md
docs/operator-runbook.md
Report vulnerabilities privately:
See SECURITY.md for responsible disclosure.
MIT License
Clone the repository:
git clone https://github.com/sentinelSCA/sentinel.git
cd sentinel
Start Sentinel:
docker compose up -d
Open the dashboard:
http://localhost:8001/dashboard
Send a test request:
curl http://localhost:8001/api/v2/security-score
You now have Sentinel running locally.
An AI agent attempts a high-impact action:
restart_service(redis)
Sentinel evaluates the request through its governance pipeline:
- Agent identity verified
- Policy decision: REVIEW
- Action recorded in a tamper-evident ledger
- Replay available for forensics
Example API response:
{
"decision": "review",
"risk": "high",
"risk_score": 0.90
}
Sentinel is inviting a small number of developers to participate in the Sentinel Founding Developer Program.
Participants receive:
• Free access to the Sentinel platform
• Early access to new features
• Direct communication with the founder
• Opportunity to influence the roadmap
We are looking for developers working with:
• AI agents
• DevOps automation
• infrastructure bots
• autonomous workflows
Participants will test Sentinel with real workflows and provide feedback.
If you are interested: