Skip to content

Security: Creed-Space/VCP-SDK

Security

SECURITY.md

Security Policy

Supported Versions

Version Supported
Python SDK 1.0.x Yes
Rust SDK 0.1.x Yes
WebMCP SDK 0.1.x Yes
VCP Specification 1.1 Yes
VCP Specification 1.0 Security fixes only

Reporting a Vulnerability

The VCP-SDK team takes security seriously. If you discover a security vulnerability, we appreciate your responsible disclosure.

How to Report

Do NOT open a public GitHub issue for security vulnerabilities.

Instead, please report vulnerabilities via one of the following channels:

  1. Email: Send a detailed report to security@creedspace.com
  2. GitHub Security Advisories: Use the private vulnerability reporting feature

What to Include

Please include the following in your report:

  • Description of the vulnerability and its potential impact
  • Affected component (Python SDK, Rust SDK, WebMCP SDK, specification, schemas)
  • Steps to reproduce the issue
  • Proof of concept code or payload, if applicable
  • Suggested fix, if you have one

Response Timeline

Stage Timeframe
Acknowledgment Within 48 hours
Initial assessment Within 5 business days
Fix development Depends on severity
Public disclosure After fix is released, coordinated with reporter

Severity Classification

Severity Description Example
Critical Remote code execution, signature bypass Forged bundle accepted as valid
High Authentication bypass, data exfiltration Private context fields leaked
Medium Denial of service, information disclosure Malformed token causes crash
Low Minor issues with limited impact Verbose error messages

Scope

The following are in scope for security reports:

  • Cryptographic signature verification (bundle signing, hash integrity)
  • Identity token parsing and validation
  • Privacy level enforcement (public/consent/private boundaries)
  • Transport layer security
  • Schema validation bypass
  • Dependency vulnerabilities in SDK packages

Recognition

We are grateful to security researchers who help keep VCP safe. With your permission, we will acknowledge your contribution in our release notes and security advisories.

Security Design

VCP's security model is built on:

  • Ed25519 signatures for bundle authentication
  • SHA-256 content hashes for integrity verification
  • Three-tier privacy architecture (public/consent/private)
  • Verify-then-Inject pattern — verification at the orchestration layer, not the LLM
  • Deterministic hooks — constitutional rules enforced outside probabilistic model behavior

For details, see Section 13 (Security Considerations) of the VCP Specification.

There aren’t any published security advisories