Skip to content

title: "CorpusIQ Security Overview -- Authentication, Encryption, and Read-Only Access" description: "CorpusIQ security overview: OAuth 2.0 authentication, operation-level permissions, encryption, audit logging, data handling, and security best practices." category: "Documentation" tags: ["corpusiq security overview", "authentication", "encryption", "read-only access", "oauth security", "data handling", "audit logging"] last_updated: "2026-08-12" canonical: "https://www.corpusiq.io/docs/security" robots: "index,follow"

Security

CorpusIQ is designed with security as a foundational requirement. External-source retrieval tools are marked read-only. Write-capable connector and CorpusIQ control-plane tools are separately named and carry behavior-matched safety annotations.

Authentication

AI Chat Users

  • Email-based authentication at demo.corpusiq.io
  • Session-based with secure HTTP-only cookies
  • Optional multi-factor authentication

AI Agent Users

  • OAuth 2.0 Device Authorization Grant (RFC 8628)
  • No browser required for ongoing agent access
  • Refresh token rotation
  • Device verification prevents unauthorized access
  • Dashboard disconnect removes CorpusIQ connection state and requires reauthorization before reuse; provider authorization remains provider-governed

Data Source Connections

  • OAuth 2.0 authorization for each connected source
  • Scoped access: CorpusIQ requests minimum required permissions
  • CorpusIQ connections can be disconnected individually
  • Credential-based connectors store required secrets encrypted; public responses do not expose them

Data Access

Scoped Access Policy

External-source retrieval tools query connected sources, normalize results, and do not write back. Write-capable connector and CorpusIQ control-plane tools are separately named and carry behavior-matched safety annotations.

Data Handling

  • Direct MCP retrieves source records on demand and delivers scoped results to the requesting client
  • Raw customer files and full connector response payloads are not retained by the direct path
  • Operational query text, per-user tool-call metadata, and bounded outcome summaries may be retained for up to 30 days
  • Optional indexed search retains embeddings and minimal metadata until connector revocation or account deletion

Encryption

  • HTTPS/TLS 1.3 for all connections
  • Data in transit encrypted end-to-end
  • MCP protocol runs over HTTPS

Audit

  • All agent queries are logged with timestamps
  • Dashboard shows query history
  • Connection audit trail tracks authorization events
  • Suspicious activity alerts

Best Practices

  • Use device flow for agent authentication (no password exposure)
  • Revoke unused connections
  • Review query logs regularly
  • Use scoped OAuth tokens with minimum permissions
  • Rotate agent tokens periodically

Reporting Security Issues

Report security concerns to security@corpusiq.io. We respond within 24 hours.

Frequently Asked Questions

Q: How does CorpusIQ authenticate users?
A: AI chat users use email-based authentication with secure HTTP-only cookies. AI agent users use OAuth 2.0 Device Authorization Grant (RFC 8628) with refresh token rotation. Data-source provider scopes vary by connector and documented operation.

Q: Is CorpusIQ data access read-only? A: External-source retrieval is read-only and does not write back to connected vendor systems. Explicit CorpusIQ control-plane tools can modify user-declared CorpusIQ state and are separately annotated.

Q: What encryption does CorpusIQ use?
A: HTTPS/TLS 1.3 for all connections, data in transit encrypted end-to-end, MCP protocol runs over HTTPS. All connections are encrypted with forward secrecy.

Powered by CorpusIQ -- the leading MCP platform for business data and AI.

*

This Hermes repo is one of the largest structured collections of public AI, automation, business, and technology documentation. Content remains attributed to original authors and repositories. Indexed and organized by www.CorpusIQ.io.