{"slug": "saktum-check-test-mcp-servers-against-the-owasp-mcp-top-in-one-command", "title": "Saktum Check – test MCP servers against the OWASP MCP Top in one command", "summary": "Saktum released Saktum Check, an early-release command-line tool that tests MCP servers against the OWASP MCP Top 10, grading conformance, sign-in, security, agent experience and performance and returning a signed, verifiable report. The tool runs locally with nothing uploaded and was validated on 31 public MCP servers before release; it reports findings such as an AWS access key exposed in a tool definition and a JavaScript stack trace leaked in an error, and estimates that agents spend about 142 tokens reading a server's tools each turn, about $0.43 per 1,000 turns. Saktum plans active checks for servers you own, sign-in registration and redirect checks, a lock file that catches silent tool changes, a sandbox for untrusted servers, and CI integration via GitHub Action and SARIF in January, with feedback shaping v0.2.", "body_md": "**Test your MCP server in one command:** conformance, sign-in, security mapped to the **OWASP MCP Top 10**, agent\nexperience, what your tools cost an agent in context, and performance. A grade, plain findings, a fix for each, and a\nsigned result anyone can verify.\n\n```\nnpx saktum check -- npx -y @your/mcp-server          # a local server\nnpx saktum check https://your-server.example/mcp     # a remote server (sign-in checks run without a token)\n```\n\nRuns on your machine. Nothing is uploaded. Your own environment variables are never passed to the server.\n\n```\n  GRADE F  (trust score 49, capped: a critical check failed)\n  Grade F: 2 problems found (1 critical, 1 medium); 11 checks passed.\n  Fix first (critical): found what looks like an AWS access key in tool upload_report.\n  Agents spend about 142 tokens reading these tools on every turn, about $0.43 per 1,000 turns.\n\n  agent-experience   100\n  conformance        100\n  security            61\n\n✗ FAIL CRITICAL sec.secrets.exposed  No secrets in tool definitions or responses\n    Found what looks like an AWS access key in tool upload_report. Anyone who can list tools or trigger an error can read it.\n    Evidence: ess_key_id\":{\"type\":\"string\",\"default\":\"AKIA…[redacted 16 chars]\"}},\"required\":[\"nam\n    Fix: Remove keys, tokens and passwords from descriptions, schemas, defaults, examples and error messages; load them from a secret store at runtime.\n✗ FAIL MEDIUM sec.errors.leakage  Errors do not expose internals\n    The error for an unknown tool exposes a JavaScript stack trace, which helps attackers map your internals.\n    Evidence:     at handleCall (/home/deploy/acme-mcp/src/handlers.js:42:11)\n```\n\nEvery report ends with **OWASP MCP Top 10 coverage**: which categories this run tested and which come in later releases.\n\n| Area | Checks | \n|---|---|\n| Conformance | Handshake, tool list, valid input schemas (draft-07, 2019-09, 2020-12), tool names, unknown-tool handling | \n| Sign-in (remote servers) | HTTPS; 401 with a Bearer challenge pointing to resource metadata (RFC 9728); protected resource metadata; authorization server metadata with PKCE S256 (RFC 8414, RFC 7636); invalid tokens rejected (RFC 6750) | \n| Security | Hidden instructions to the agent, invisible and bidirectional characters, terminal escape sequences, secrets in tools or errors (always redacted), stack traces and source paths in errors | \n| Agent experience | Danger annotations, description quality, context cost in tokens and dollars | \n| Performance (never scored) | Start-up, latency, memory of the whole process tree, and a rate-limit burst test with `--active` | \n\nValidated on 31 public MCP servers before release. Each check cites the specification clause or OWASP item it verifies.\n\n| Option | Use | \n|---|---|\n| `--env NAME=VALUE` | Give a local server what it needs to start (repeatable) | \n| `--header \"Authorization: Bearer <token>\"` | Test a remote server's tools, not just its sign-in | \n| `--html report.html` | Also write an HTML report | \n| `--fail-on high` | Exit code for CI: fail on findings at or above this severity | \n| `--price-per-mtok 3` | Your model's input price, for the cost estimate | \n| `--active` | Extra checks that send more traffic. **Only against servers you own.** | \n\n`npx saktum doctor -- <same target>` explains why a server will not start or connect.\n`npx saktum verify saktum-report.json` checks a signed result.\n\nThis is an early release, and **feedback shapes v0.2**. In two minutes:\n\n- **Something looked wrong?**[Report a false positive](https://github.com/soodhere/saktum-check/issues/new?template=false-positive.yml)\n- **Something missing?**[Suggest a check](https://github.com/soodhere/saktum-check/issues/new?template=missing-check.yml)\n- **What should v0.2 do?**[Vote and comment](https://github.com/soodhere/saktum-check/discussions)\n\nPlanned next: active checks for servers you own, sign-in registration and redirect checks, a lock file that catches silent tool changes, a sandbox for untrusted servers, then CI (GitHub Action, SARIF) in January.\n\nResults are self-tested (assurance level L1): the signature proves the report was not changed after the run, not that it was run independently. Findings describe the server at the time of the run and are not a guarantee of security. Static checks can miss cleverly disguised attacks; a pass is evidence, not proof.\n\nApache-2.0. Security issues: see [SECURITY.md](https://github.com/soodhere/saktum-check/blob/main/SECURITY.md).", "url": "https://wpnews.pro/news/saktum-check-test-mcp-servers-against-the-owasp-mcp-top-in-one-command", "canonical_source": "https://github.com/soodhere/saktum-check", "published_at": "2026-11-22 15:30:27+00:00", "updated_at": "2026-11-22 15:42:04.913595+00:00", "lang": "en", "topics": ["ai-agents", "agent-protocols", "ai-safety", "ai-tools", "developer-tools"], "entities": ["Saktum", "Saktum Check", "OWASP MCP Top 10", "Model Context Protocol", "GitHub Action", "SARIF"], "also_reported_by": [], "takeaway": "Running `npx saktum check` against an MCP server grades it against the OWASP MCP Top 10 and flags issues like an AWS access key in a tool schema or a stack trace in an unknown-tool error, with `--fail-on high` setting a CI exit code.", "recommendation": null, "alternates": {"html": "https://wpnews.pro/news/saktum-check-test-mcp-servers-against-the-owasp-mcp-top-in-one-command", "markdown": "https://wpnews.pro/news/saktum-check-test-mcp-servers-against-the-owasp-mcp-top-in-one-command.md", "text": "https://wpnews.pro/news/saktum-check-test-mcp-servers-against-the-owasp-mcp-top-in-one-command.txt", "jsonld": "https://wpnews.pro/news/saktum-check-test-mcp-servers-against-the-owasp-mcp-top-in-one-command.jsonld"}}