Understand what each tool can do—and what it can change.

Analyze MCP servers, OpenAPI specifications, agent tools, workflow actions, and callable capabilities to identify authentication requirements, permissions, side effects, trust-boundary crossings, approval needs, destructive operations, and excessive scope before launch.

Analyze Tool Definitions View Sample Tool Analysis

WHAT CAN YOUR AI TOOLS REALLY DO?

Discover

Inventory callable capabilities from MCP, OpenAPI, agent frameworks, plugins, workflow systems, and custom tool schemas.

Classify

Classify authentication, read, write, delete, send, execute, administer, browser, filesystem, secret, network, and external-effect capabilities.

Identify control gaps

Find excessive scope, missing approvals, destructive operations, risky side effects, inherited permissions, and cross-boundary capabilities.

Hand off

Contribute normalized tool context and findings to Threat Canvas, Authority Graph, Agent Permission Analysis, Agent Authority Review, and release decisions.

Core capabilities

What Tool Analyzer does.

Tool Inventory

Build an inventory of callable tools, operations, workflow actions, MCP servers, API capabilities, and agent-accessible integrations.

Capability Classification

Classify each operation by the data it can access, the action it can perform, the identity it uses, the side effects it can cause, and whether the action is reversible.

Authentication and permission model

Map credentials, API grants, inherited scopes, runtime identities, argument controls, permission boundaries, and side effects before deployment.

Trust Boundary Mapping

Identify operations that cross trust, tenant, identity, data, network, or external-effect boundaries and determine where approvals or stronger controls are required.

Connected analysis handoff

Produce structured tool capsules and findings that contribute to Threat Canvas, Authority Graph, Agent Permission Analysis, Agent Authority Review, and Attack Path Analysis.

Static Analysis First

Analyze definitions, schemas, and approved implementation evidence without invoking production tools, changing customer systems, or claiming observed runtime behavior.

Evidence & signals

What you get out of the box.

Security Model

Tool inventory
Tool capsules
Capability classification
Authentication and permission model
Side-effect map
Trust-boundary crossings
Approval requirements
Evidence references

Deliverables

Threat Canvas inputs
Authority Graph inputs
Permission findings
Control recommendations
Release-review signals
Engineering backlog

What it produces

Structured tool context for modeling, review, and hardening.

A tool capsule is the normalized Workbench representation of a callable tool’s schema, authentication, permissions, side effects, boundaries, and evidence references.

Tool inventoryTool capsulesCapability classificationAuthentication modelPermission scopeSide effectsTrust-boundary findingsApproval requirements

AI SECURITY WORKBENCH

Ready to make callable tool authority reviewable?

Use Tool Analyzer to inventory callable capabilities, normalize their authority and side effects, and carry that context into Threat Canvas, Authority Graph, hardening, and release review.

Scope an AI Security Review See the full Workbench

Continue through the Workbench

Threat Canvas

Place tools, authentication, data movement, and trust-boundary crossings in the system model.

Explore

Continue through the Workbench

Authority Graph

Analyze how tools compose with identities, agents, credentials, approvals, and downstream actions.

Explore

Continue through the Workbench

Agent Permission Analysis

Apply deterministic security rules to agent and tool configurations.