Skip to content

Latest commit

 

History

History

README.md

Commune Security Documentation

Commune provides multi-layered security that protects both outbound sending reputation and inbound processing integrity — all enabled by default, zero configuration required.

These docs explain each security layer, what it does, and how developers and agents interact with it.


Security Layers

Outbound (Sending) Protection

Layer Doc What It Does
Email Validation email-validation.md Validates every recipient before sending — catches spam traps, disposable addresses, invalid domains
Bounce Intelligence bounce-intelligence.md Classifies hard vs soft bounces, auto-suppresses, self-heals when addresses recover
Reputation Autopilot reputation-autopilot.md Real-time circuit breaker — auto-pauses sending before ISPs flag you
Domain Warmup domain-warmup.md Graduated sending schedule for new domains to build ISP trust safely
Deliverability Compliance deliverability-compliance.md One-click unsubscribe (RFC 8058), DMARC monitoring, Gmail/Yahoo compliance

Inbound (Receiving) Protection

Layer Doc What It Does
Spam & Phishing Detection inbound-threat-detection.md Multi-signal spam scoring, phishing detection, mass attack blocking
AI Agent Safety ai-agent-safety.md Prompt injection detection for AI agents processing email
Attachment Scanning attachment-scanning.md File type blocking, magic byte inspection, malware signatures, ClamAV

Webhook & Delivery Infrastructure

Layer Doc What It Does
Webhook Delivery Guarantee webhook-delivery-guarantee.md Retries, exponential backoff, circuit breaker, dead letter queue, delivery tracking API
Webhook Signature Verification webhook-signatures.md HMAC-SHA256 signing so agents can verify webhook authenticity

Infrastructure Hardening

Layer Doc What It Does
Encryption & Hardening encryption-and-hardening.md AES-256-GCM encryption at rest, injection prevention, rate limiting, error redaction

For Developers

Every security feature is on by default. You don't configure or enable anything. When you send an email, validation and reputation protection happen automatically. When you receive an email, threat detection and AI safety scanning happen before your webhook fires.

Security metadata is included in every webhook payload and API response, so your application always knows the risk profile of what it's processing.

For AI Agents

Commune is purpose-built for AI agent email infrastructure. The security layers are designed to:

  • Protect agents from manipulation — Prompt injection detection flags emails that try to hijack agent behavior
  • Guarantee webhook delivery — Retries with exponential backoff ensure your agent never misses an email, even during downtime
  • Provide rich context — Every email arrives with spam scores, phishing flags, injection risk levels, and attachment scan results so agents can make informed decisions
  • Verify authenticity — HMAC signatures on every webhook let agents cryptographically verify that payloads came from Commune and weren't tampered with