#2323rd of 40 in Security
cybermes
Bug bounty automation with a Go MCP server and 200+ security playbooks
- Stars
- 1k
- License
- Apache-2.0
- Last commit
- Nov 2026
- Last release
- Nov 2026
- Language
- Python
20 on 19 Aug 20261,044 on 18 Nov 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Cybermes runs authorized recon and vulnerability research either as a standalone CLI agent (Hermes) or as a Go MCP server (cybermes-mcp) for AI editors such as Cursor, Claude Desktop and Cline. It orchestrates subfinder, httpx, katana, ffuf, nuclei and sqlmap, filters their output with a Go stream filter, and requires a Python PoC script before a finding is logged. Results are compiled into SUMMARY.md, metadata.json, HTML and PDF reports.
Who it is for: Bug bounty hunters and pentesters who want AI-driven recon and reporting
Strengths
- Two modes: autonomous CLI agent, or MCP server for existing AI editors
- Offline knowledge base of 200+ playbooks plus PayloadsAllTheThings and HackTricks data
- smart_pipe filter archives raw tool output and passes only high-signal lines to the model
- Per-target workspaces with PoC scripts, evidence and PDF/HTML/Markdown reports
Weaknesses
- Depends on external tools (subfinder, httpx, katana, ffuf, nuclei, sqlmap) installed separately
- 'Zero-false-positive' claim rests on a PoC gate; no benchmark is given
- README is heavy on sponsor and donation promotion, including an affiliate API link
What it needs
- Docker + Compose
- Needs Go 1.22+, Python 3.11+, subfinder, httpx, katana, ffuf, nuclei, sqlmap, Node.js (npx) for cybermes-mcp
What changed
- v3.5.1 ↗ (opens in a new tab)Security fix
“Cybermes v3.5.1 introduces an automated update system for the NPM package, native Go Nmap service fingerprinting, finding deduplication in report aggregators, hardened scope guards, and security fixe…”
- v3.5.0 ↗ (opens in a new tab)Security fix
“Target Slug Sanitization: Enforced sanitizeSlug() across all MCP tool handlers (toolsrecon, toolsreports, toolsfuzz, toolsnuclei) using regex [^a-z0-9]+ normalization. Prevents directory traversal at…”
From the release notes. No flag means the note does not say so, not that the upgrade is safe.
Also in Security
See all 40| Rank | Project | Score |
|---|---|---|
| 1 | shannonAutonomous pentester that exploits web apps and APIs using source analysis | 92 out of 100 |
| 2 | ifixaiPython CLI that audits AI agents for policy, access and reliability failures | 87 out of 100 |
| 3 | StrixAutonomous multi-agent pentesting CLI that validates vulnerabilities with working PoCs | 85 out of 100 |
| #4 | ai-infra-guardRed-teaming platform for AI infra CVEs, MCP servers, agent skills and jailbreaks | 84 out of 100 |
| #5 | decepticonAutonomous red-team agents that run multi-stage attack chains from an approved plan | 84 out of 100 |