convez.ai
MCP Server
MukundaKatta

secretsniff-mcp

MukundaKatta/secretsniff-mcp

Ver no GitHub
InicialPluginScore · de 0 a 100

MCP server: scan text/code for accidentally-committed secrets. AWS/GitHub/Slack/Stripe keys, JWTs, RSA keys, high-entropy strings.

Instalação

$npx convez-ai MukundaKatta/secretsniff-mcp

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
0
Forks
0
Acompanhando
0
Saúde
42%
Último commit
há 2 meses
Versão
—
Issues abertas
4
Contribuidores
1

README

A tradução para o português deste README ainda está na fila.

secretsniff-mcp

npm mcp license

An MCP server that gives AI assistants the ability to scan text or code for accidentally-committed secrets before they ship.

Works with Claude Desktop, Cursor, Cline, Windsurf, Zed, and any other MCP client.

What it catches

DetectorPattern
AWS_ACCESS_KEYAKIA… 20-char keys
GITHUB_TOKENghp_/gho_/ghu_/ghr_/ghs_…
SLACK_TOKENxox[baprs]-…
STRIPE_KEYsk_live_/sk_test_/pk_live_/pk_test_…
JWTthree url-safe base64 segments
RSA_PRIVATE_KEY-----BEGIN RSA PRIVATE KEY----- marker
SSH_PRIVATE_KEY-----BEGIN OPENSSH PRIVATE KEY----- marker
GENERIC_API_KEYapi[_-]?key\s*[=:]\s*"…" style assignments
HIGH_ENTROPYstrings ≥ 32 chars with Shannon entropy ≥ 4.5 bits/char

Tools exposed

scan_text

Scan a string for secrets. Returns one entry per finding with line/column, the matched substring, and Shannon entropy.

{ "text": "const KEY = 'AKIAIOSFODNN7EXAMPLE';" }

→

{
  "findings": [
    {
      "kind": "AWS_ACCESS_KEY",
      "line": 1,
      "column": 14,
      "start": 13,
      "end": 33,
      "matched": "AKIAIOSFODNN7EXAMPLE",
      "entropy": 3.78
    }
  ]
}

scan_file

Read a file from disk and scan it. Same return shape as scan_text plus the file path. Useful when an agent has just written a file and wants to double-check it before committing.

Configure your MCP client

Claude Desktop's claude_desktop_config.json:

{
  "mcpServers": {
    "secretsniff": {
      "command": "npx",
      "args": ["-y", "@mukundakatta/secretsniff-mcp"]
    }
  }
}

Cursor / Cline / Windsurf / Zed configs follow the same command + args pattern.

Why this is not a --lint flag

This is meant for the agentic loop: an LLM that just wrote a config or edited a file calls scan_text on its own output before declaring success. Existing CLI scanners (gitleaks, trufflehog) target the human-driven pre-commit workflow; they're great at that and I'd still run them in CI. secretsniff-mcp covers the gap during the LLM's editing turn.

License

MIT.

Ver o original no GitHub

Bom para

Relacionados