Skip to content

fix(security): resolve MCP registry security audit findings (Python CVEs, SIEM endpoint, dynamic ledger salt) - #80

Merged
divyanshujethi merged 2 commits into
mainfrom
fix/security-audit-remediation
Nov 19, 2026
Merged

divyanshujethi merged 2 commits into
mainfrom
fix/security-audit-remediation

Conversation

@divyanshujethi

Copy link
Copy Markdown
Collaborator

Summary of Security Remediations

Directly resolves all 15 audit findings reported on MCP Marketplace:

  • Bumped Python dependencies (langchain-core>=0.3.74, langgraph>=0.2.70, pytest>=8.3.4) to eliminate all 7 supply chain CVEs
  • Removed hardcoded vendor telemetry fallback endpoint from siem test; requiring explicit opt-in
  • Removed placeholder API key and enforce real credentials for Datadog / Splunk
  • Pinned exact versions for delegated packages (@ritualdev-lab/dbshield@0.2.1, @ritualdev-lab/skillprism@0.1.0)
  • Added stdin/env token support and CLI argument exposure warning in login`n- Replaced hardcoded ledger salt with dynamic per-installation cryptographically secure random key (~/.agentshield/ledger.key with 0600 mode)
  • Created standalone untruncated MCP entry point (bin/mcp.js registered as agentshield-mcp)
  • Scoped LAUNCHGUIDE.md cleanly to npm package

…VEs, opt-in SIEM URL, dynamic ledger salt, pinned packages, standalone MCP bin)
@github-actions

Copy link
Copy Markdown

👋 Thanks for opening this Pull Request, @divyanshujethi!

Thank you for contributing to AgentShield! 🛡️

Here is what happens next:

  1. 🧪 Automated Security Gate: Our CI workflow (pr-gate.yml) audits your diff for accidental credential exposure and runs test suites.
  2. 🔍 Code & Security Review: A maintainer from @RitualDev-Lab will review your changes shortly.
  3. 🌟 Contributor Recognition: Once merged, you'll be featured on our contributors list!

💡 Tip: Make sure secrets/mock API keys in tests are synthetic/redacted and tests pass locally (pnpm test).

@github-actions

github-actions Bot commented Nov 19, 2026 •

Copy link
Copy Markdown

🛡️ AgentShield Pull Request Security Audit

✅ Security Gate Passed — No Exposed Secrets Detected

  • Files Audited: 10
  • Commit Evaluated: d0208b8
  • Status: Diff is clean. Zero-leak credential policy verified for autonomous AI & developer tools.

AgentShield: OWASP LLM06 Compliant


Protected by AgentShield — Zero-Leak Security for AI Agents & Engineering Teams.

@divyanshujethi
divyanshujethi merged commit 77250ac into main Nov 19, 2026
11 checks passed
@divyanshujethi
divyanshujethi deleted the fix/security-audit-remediation branch November 19, 2026 15:34
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant