From Model Output to Business Action
A security briefing on agents that can read, write, send, transact, deploy, delete, or operate business systems.
Request this briefingEducational resources
Practical AI agent security resources.
Plain-language materials for security, engineering, governance, procurement, enterprise sales, and board teams reviewing agents that can reach data or take action.
Security and engineering
Start with questions you can verify.
Use these materials to clarify scope, identify evidence, align teams, set architecture requirements, and prepare a focused assessment.
40 checks / 8 control domains
Printable checklist
AI Agent Exposure Checklist
Inventory, ownership, authority, tools, MCP, data, approvals, evidence, containment, and governance.
Open checklistidentity → gateway → sandbox → evidence
Engineering field guide
MCP Security Field Guide
A defensive baseline for tool servers and agent integrations that can reach enterprise systems.
Read the guideevidence > assertions
Due diligence template
AI Agent Security Questionnaire
Questions enterprise buyers should ask vendors whose agents access customer data or take action.
Open questionnaireidentity · authority · action · evidence
Plain-language reference
AI Agent Security Glossary
Shared definitions for agents, tool calls, MCP, delegated authority, approval, containment, and evidence.
Open glossaryExecutive alignment
Frame agentic risk in business terms.
Autonomy is not inherently unsafe. Risk increases when authority is broad, oversight is unclear, evidence is incomplete, or containment has not been tested.
the new non-human insider
Board brief
AI Agents: From Output Risk to Action Risk
Ten board-level questions, a control model, and meaningful metrics for oversight.
Open board briefmap → prioritize → control → prove
Service overview
AI Agent Exposure Review
Review the available fixed-scope service, deliverables, assessment domains, engagement boundaries, process, and public starting estimate.
Review the engagementkeep AI on mission.
Media and brand kit
Fugitive Intelligence Press Kit
Company boilerplate, slogans, launch copy, brand colors, and downloadable local logo assets.
Open press kitBriefings and workshops
Use one shared control language.
Private sessions can be tailored to agreed workflows, architecture, customer questions, and risk decisions. They are educational and do not replace legal, audit, or compliance advice.
MCP Tool-Call Security
A technical session on identity, token handling, tool integrity, execution isolation, egress, approvals, and evidence.
Request this briefingBoard Oversight for AI Agents
An executive discussion on delegated authority, high-impact actions, accountable approval, evidence, and response readiness.
Request this briefingUse the materials internally
Share, print, and adapt the questions to your scope.
These resources are intended to accelerate internal security work and buyer conversations. Validate the questions against your architecture and obligations. Attribution is appreciated when redistributed outside your organization.
- No registration wall or marketing tracker
- Clean print layouts for internal workshops
- Plain-language definitions for technical and executive audiences