Security Measures in codebase-memory-mcp: VirusTotal, SLSA Level 3, and Sigstore Signatures Explained

codebase-memory-mcp implements a seven-layer security pipeline including SLSA Level 3 provenance, Sigstore cosign signatures, VirusTotal scans across 70+ engines, and SHA-256 checksum verification to ensure binary integrity and authenticity.

The codebase-memory-mcp repository by DeusData adopts a comprehensive security framework that protects both the build process and distributed binaries. This MCP (Model Context Protocol) server for codebase memory implements cryptographic provenance, automated malware scanning, and multi-factor verification to guarantee that every release is authentic and untampered.

Build-Time Provenance with SLSA Level 3 Attestation

Every release archive is generated inside a trusted GitHub Actions workflow that produces a cryptographic SLSA Level 3 attestation. According to SECURITY.md, this provenance ensures the binary was built from the exact source code in the repository without tampering. The trusted workflow file .github/workflows/_build.yml acts as the secure builder for all releases.

Verifying SLSA Provenance

Users can verify the attestation using the GitHub CLI to confirm the binary originated from the official build workflow:

gh attestation verify codebase-memory-mcp-linux-amd64.tar.gz \
  --repo DeusData/codebase-memory-mcp \
  --signer-workflow DeusData/codebase-memory-mcp/.github/workflows/_build.yml

Sigstore Cosign Signatures for Keyless Verification

All release artifacts are signed keylessly using Sigstore's cosign. The signature bundle is published alongside each release artifact, allowing validation without managing private keys. As documented in the Security table of README.md, this provides transparent, public-key infrastructure for artifact authentication.

Verifying Cosign Signatures

To validate the signature bundle against the downloaded archive:

cosign verify-blob --bundle codebase-memory-mcp-linux-amd64.tar.gz.bundle \
  codebase-memory-mcp-linux-amd64.tar.gz

VirusTotal Scanning Across 70+ Antivirus Engines

Before any draft release is promoted to public, binaries undergo mandatory VirusTotal scanning. The SECURITY.md file confirms that detection from any of the 70+ antivirus engines blocks the release pipeline. Scan results are published in release notes with specific file hashes showing 0/72 detections for all platforms.

Platform-Specific Scan Results

The repository provides direct links to VirusTotal results for each platform (v0.7.0 example):

  • Linux amd64: 0/72 detections
  • Linux arm64: 0/72 detections
  • macOS arm64: 0/72 detections
  • macOS amd64: 0/72 detections
  • Windows amd64: 0/72 detections

Users can verify scans manually by visiting https://www.virustotal.com/gui/file/<file-hash>/detection with the appropriate SHA-256 hash.

SHA-256 Checksums and Install Verification

A checksums.txt file is generated for every release. The scripts/setup.sh install script automatically validates these checksums before extraction, while scripts/license-gate-check.py performs additional validation to ensure license and checksum integrity match expectations.

Manual Checksum Verification

sha256sum -c checksums.txt

This protects against tampered downloads by ensuring the file hash matches the cryptographically signed release.

Static Analysis and Security Auditing

The repository implements continuous security scanning through CodeQL static analysis (SAST) on every push. Any open alerts block the release pipeline. Additionally, an 8-layer security audit suite runs during CI covering:

  • Static allow-list validation
  • Binary string checks
  • Network egress monitoring
  • Install-path validation
  • Smoke-test hardening
  • UI domain checks
  • MCP robustness testing
  • Vendored-dependency integrity verification

These measures are detailed in the Security Measures section of SECURITY.md.

Runtime Security Safeguards

The distributed binary operates with privacy-first defaults: no telemetry is collected. The application only performs a non-blocking update check to GitHub with bounded timeouts (curl --max-time 5), ensuring minimal external exposure while maintaining security patch awareness.

Summary

  • SLSA Level 3 provenance ensures builds originate from trusted GitHub Actions workflows defined in .github/workflows/_build.yml
  • Sigstore cosign provides keyless cryptographic signatures for all release artifacts
  • VirusTotal scans across 70+ engines block releases containing potential malware
  • SHA-256 checksums in checksums.txt enable integrity verification before installation
  • CodeQL static analysis and an 8-layer security audit suite prevent vulnerable code from reaching releases
  • Runtime safeguards ensure local execution without telemetry or unbounded network access

Frequently Asked Questions

How do I verify that a codebase-memory-mcp binary is authentic?

Verify authenticity through three independent checks: run gh attestation verify to confirm SLSA Level 3 provenance from the trusted workflow, use cosign verify-blob to validate the Sigstore signature bundle, and check sha256sum -c checksums.txt to ensure file integrity. All three validations must pass to confirm the binary was built from the official source and has not been tampered with.

What happens if VirusTotal detects malware in a codebase-memory-mcp release?

If any of the 70+ antivirus engines on VirusTotal report a detection, the release pipeline automatically blocks promotion from draft to public status. This prevents compromised binaries from reaching users, and the security team investigates the detection before any release proceeds.

Does codebase-memory-mcp collect telemetry or phone home during runtime?

No. The binary runs locally without collecting telemetry. It only performs a non-blocking check for newer releases via GitHub's API with a strict 5-second timeout (curl --max-time 5), ensuring users remain informed of security updates without sacrificing privacy.

What is the 8-layer security audit suite in codebase-memory-mcp?

The 8-layer audit suite runs during continuous integration and includes static allow-list checks, binary string validation, network egress monitoring, install-path verification, smoke-test hardening, UI domain validation, MCP robustness testing, and vendored-dependency integrity checks. This comprehensive scanning ensures release artifacts meet strict security standards before publication.

Have a question about this repo?

These articles cover the highlights, but your codebase questions are specific. Give your agent direct access to the source. Share this with your agent to get started:

Share the following with your agent to get started:
curl -s "https://instagit.com/install.md"

Works with
Claude Codex Cursor VS Code OpenClaw Any MCP Client

Maintain an open-source project? Get it listed too →