LV2 · Toolooo Standard
MCP Security Doctor
Review MCP tool and server JSON for risky capabilities, broad permissions, approval gaps and sensitive data signals.
Is this MCP definition granting more power than its task needs? Paste JSON, review the signals, then verify the real safeguards. Nothing is executed or connected.
100,000 UTF-8 bytes · 5,000 values · 24 levels · 100 tools. Inputs stay in page memory. Results hide values and unknown property names; your pasted editor remains visible.Try a fictional example
Accepts one tool, a tool array, a tools/list response, mcpServers configuration or generic tool JSON. It recognizes common fields, not every client format. Results appear only after analysis.
What this review cannot establish
This is not malware detection, penetration testing, runtime scanning or a security certification. It does not resolve schema references, inspect server code or prove that an approval, read-only hint, allowlist or output schema is enforced. No result copy/export is offered: findings and configurations stay off shared links and analytics.
MCP tool definitions and trust boundaries · MCP security best practices
What this tool does and why
Before you start
What
Review MCP tool and server JSON for risky capabilities, broad permissions, approval gaps and sensitive data signals.
Why
An agent tool can turn untrusted instructions into real side effects. Reviewing its declared power, resource scope and approval boundaries helps identify what must be verified before enabling it.
Your Result and Action update in the tool as you change its inputs.
Model note
This browser-side model explains the inputs you choose. Check an important decision against the real system, brief, or production constraints.
Next
How this tool stays honest
How it works
Paste tool definitions, a tools/list response or server configuration for a bounded static security review. Inspect declared capabilities, permissions, approval requirements, schemas, suspicious instructions and trust boundaries with hidden result values. Nothing is executed, connected, uploaded or certified safe.
Assumptions
The tool runs in your browser and models the information you provide. It does not know facts outside that model.
Limitations
Static JSON heuristics only: no server connection, execution, schema-reference resolution, malware detection or security certification. Annotations and custom approval flags are declarations, not verified safeguards. Results hide values and unknown property names; the editor remains visible.
Understand MCP Security Doctor
What this tool helps you understand
Paste tool definitions, a tools/list response or server configuration for a bounded static security review. Inspect declared capabilities, permissions, approval requirements, schemas, suspicious instructions and trust boundaries with hidden result values. Nothing is executed, connected, uploaded or certified safe.
When to use it
Before granting an agent access to an MCP server or newly exposed tool.
Try a scenario
Compare a constrained project-status reader with an arbitrary-path deletion tool. Review the declared scope and human approval boundary; annotations do not prove enforcement.
How to read the result
Review FAIL signals first, then WARN in context. Narrow scope and verify high-impact approval gates in the actual client/server. PASS applies only to the observed declaration; INFO records limits and unknowns, never proof of safety.
Assumptions and limits
Static JSON heuristics only: no server connection, execution, schema-reference resolution, malware detection or security certification. Annotations and custom approval flags are declarations, not verified safeguards. Results hide values and unknown property names; the editor remains visible.
Toolooo is part of Khizooology, created byKhizar Imtiaz (khizooo).