M
MCP Directory CURATED DIRECTORY
911 MCP servers indexed
中文
← Back to directory
COMPARE UP TO 4 SERVERS

Compare MCP servers

Compare scores, permissions, risks, and fit in one decision-focused table.

DimensionBlitz StrikeRecon at speed. Analysis in depth. Validation before report.GitHub MCP ServerGitHub's official MCP server for managing repos, issues, PRs, and workflows via natural language
FMRS42 / 100 · D76 / 100 · B
Reliability6 / 2013 / 20
Security and permissions4 / 2014 / 20
Maintenance9 / 2018 / 20
Documentation13 / 2017 / 20
Setup experience10 / 2014 / 20
Best for
  • Authorized security researchers and red/blue team engineers who want verify-before-report workflows
  • Agent users who want a full penetration-testing flow driven by one call inside an MCP client
  • Developers and security reviewers needing code audit (taint analysis, source-to-sink, variant analysis)
  • Teams that want a structured methodology plus knowledge base (manuals, playbooks, escalation chains) instead of scattered scripts
  • Teams that want an AI assistant to directly operate on GitHub repos and collaboration workflows
  • Users already in the GitHub Copilot ecosystem who want a zero-deployment remote option
Not for
  • Anyone wanting to test external systems without explicit owner authorization
  • Anyone treating a scan hit as a confirmed vulnerability and skipping STRIKE validation
  • Users expecting everything to work offline out of the box — payloads and templates need `blitzstrike sync-data`, and fofa_search needs FOFA_EMAIL/FOFA_KEY
  • Users looking for an official or vendor-endorsed product; this is a community project
  • Scenarios where you don't want the assistant to have write access to repos (enable only read-only toolsets)
  • Environments with strict network isolation for private repos that can't reach the official remote endpoint
Required permissions
  • Read the source directory or files being audited
  • Run a local process over stdio (command `blitzstrike`, args `serve --mcp`)
  • Make outbound HTTP requests to the audited target for live validation and recon
  • Write to client MCP config files (install merges into existing config, never overwrites)
  • Read/write memory and data cache under ~/.blitzstrike (overridable via BLITZSTRIKE_HOME / BLITZSTRIKE_DATA)
  • Optional: FOFA_EMAIL / FOFA_KEY for FOFA access, and H1_USERNAME to add the X-HackerOne-Research header
  • ensure_tool may auto-install missing security tools on the local machine
  • A personal access token (PAT) or OAuth App token; effective scope depends on the token's own permissions
  • Enabling toolsets like actions/issues/pull_requests grants write access — request tokens on a least-privilege basis
Risks and side effects
  • Live validation and live_recon send real requests to targets; misconfigured scope or missing authorization can cause harm or legal exposure
  • ensure_tool auto-installation modifies the local environment and introduces supply/version risk
  • Payloads, nuclei templates, and exploitation manuals are offensive material that can be misused
  • Memory is stored append-only as JSONL at ~/.blitzstrike/memory.jsonl and may accumulate sensitive target data
  • FOFA and HackerOne credentials are supplied via environment variables and can leak
  • Automated scan results can still be false positives — the README insists every finding stays a hypothesis until strike_verify confirms it
  • Write toolsets (creating/merging PRs, triggering workflows) can cause accidental changes if the token is overscoped — try a read-only toolset first
  • In hosted mode, credentials travel via the Authorization header — make sure the client-to-api.githubcopilot.com connection is trusted
Supported clientsClaude Desktop, Claude Code, Cursor, OpenCode, Hermes, Gemini, Windsurf, GitHub Copilot, ClineClaude Desktop, Claude Code, VS Code, Cursor, Windsurf, JetBrains IDEs, Zed, Amp
Tools5215