{"resourceId":"crowdstrike-agent-graph-falcon-guardian-2026","versions":[{"version":"external-d893fcfbc32769abd54dcee8d71d1acd375113551006db25d13c282d8efec19a","resource":{"id":"crowdstrike-agent-graph-falcon-guardian-2026","title":"CrowdStrike Agent Graph Shows How Agents Act on the Endpoint","organization":"CrowdStrike","sector":"Cybersecurity vendor; AI agent security (AIDR)","geography":"Global","publishedAt":"2026-10-30","publicationDate":"2026-10-30","eventDate":null,"sourceName":"CrowdStrike","sourceLabel":"CrowdStrike blog (Securing AI)","sourceUrl":"https://www.crowdstrike.com/en-us/blog/crowdstrike-agent-graph-shows-how-agents-act-on-endpoint/","evidenceClass":"vendor-claim","outcomeClass":"emerging","topics":["developers-agents","data-security","governance-procurement"],"finding":"CrowdStrike describes Agent Graph, the relational layer under Falcon Guardian, which links each AI agent to users, hosts, models, prompts, tools, skills, MCP servers, processes, files and detections. It powers three views: an inventory of sanctioned and shadow agents, a session graph tracing prompt to tool calls to endpoint actions, and a 0-100 agent threat score. Agent hooks, which can also check prompts or tool calls against policy and block actions, are available for Claude Code, OpenAI Codex, GitHub Copilot, OpenCode and Cursor.","sledRelevance":"Universities, state IT shops and research labs increasingly host coding agents and desktop AI assistants on managed endpoints. Agent inventory and prompt-to-action tracing address shadow-AI governance questions those institutions face, but the post gives no public-sector deployment evidence.","evidence":"Vendor product blog with an illustrative example (a pasted setup note causing an installer to run, which Guardian blocked). No customer data, accuracy figures or independent testing are reported.","architectureImplications":"Relies on the existing Falcon sensor plus endpoint network inspection and per-agent hooks; coverage depends on which agents are supported; the post does not state operating-system coverage.","governanceImplications":"An agent inventory and threat score can support AI-use policies and approvals; the hook-based blocking enforces policy at prompt and tool-call level.","securityPrivacyImplications":"Hooks capture exact prompt text, tool inputs and outputs, which may include sensitive or regulated data (student records, CJIS, PHI); retention and access to this telemetry need privacy review.","caveats":"Vendor marketing; illustrative scenario only. Supported-agent list is limited; unsupported agents get less detail. No evidence of detection accuracy or false-positive rates.","streamIds":["crowdstrike"],"roles":{"sales":"Shadow AI visibility is the framing point: Agent Graph promises an inventory of coding agents and assistants on managed endpoints and a trace from prompt to action. For university and state IT leaders drafting AI-use policies, the bounded conversation is about what agents already run and who approved them. Interpretation: propose a limited discovery pilot on a defined device group. The post offers no accuracy or customer outcome data, so avoid claims of complete coverage or measured risk reduction.","engineering":"Supported-agent scope is the framing point: hooks exist today for Claude Code, Codex, GitHub Copilot, OpenCode and Cursor, and other agents get less detail. Interpretation: pre-sales should list which agents the customer actually uses, confirm operating-system coverage with the vendor, and test whether policy checks on tool calls interfere with legitimate developer work. Prompt and tool-output capture creates sensitive telemetry, so retention, access controls and data-classification rules must be designed before broad rollout.","delivery":"Policy translation is the framing point: an inventory and threat score are only useful if mapped to an approved-agent policy and response playbook. Interpretation: delivery would pilot on a research or developer group, tune hook policies, and define who reviews high-score agents. Ownership sits with the security operations team and the AI governance lead. Proposed acceptance criteria include an agent inventory reconciled with policy and a measured false-block rate during the pilot."},"roleDetails":{"sales":{"customerProblem":"Institutions lack visibility into which AI agents run on managed endpoints and what those agents do.","stakeholders":["CISO","AI governance or policy lead","Research computing director","Developer platform owner"],"discoveryQuestions":["Which coding agents or AI assistants are in use today?","Is there an approved-agent policy?","Who would review prompt and tool-call telemetry?"],"valueHypothesis":"Agent inventory and prompt-to-action tracing could support AI-use governance; benefits unmeasured.","potentialEngagement":"A scoped discovery pilot on a developer or research device group.","unsupportedClaims":["Complete agent coverage","Detection accuracy","Public-sector deployment results"]},"engineering":{"fit":"Organizations already running the Falcon sensor with agent use on managed endpoints.","architectureIntegration":"Uses Falcon sensor, endpoint network inspection and native agent hooks; feeds Falcon Guardian views.","prerequisites":["Falcon sensor deployment","List of agents in use","Data-handling rules for prompt telemetry"],"constraints":["Hooks only for listed agents","Unmanaged or personal devices out of scope","Privacy rules for captured prompts"],"security":"Captured prompts and outputs must be protected and retention-limited.","proofOfValue":["Compare discovered agents with a manual survey","Measure false blocks of legitimate tool calls","Test detection of a planted malicious instruction"]},"delivery":{"work":"Pilot deployment, hook policy tuning and review workflow for high-score agents.","dependencies":["Approved-agent policy","Privacy review","Developer or researcher cooperation"],"ownership":"Security operations with the AI governance lead.","skillsAdoption":"Analysts must learn to read agent session graphs and threat scores.","governanceCheckpoints":["Privacy impact review","Policy approval before blocking mode","Pilot results review"],"acceptanceCriteria":["Proposed: inventory reconciled with approved-agent list","Proposed: false-block rate measured and accepted","Proposed: retention settings documented"],"risks":["Developer pushback from blocking","Sensitive data in telemetry","Gaps for unsupported agents"]}},"retrievedAt":"2026-10-30T22:12:07-05:00","enrichedAt":"2026-10-30T22:21:24-05:00","enrichmentBasis":"retrieved source","accessibilityWorkforceImplications":"Analysts need new skills to triage agent sessions; no workload figures given.","procurementImplications":"Capability is part of Falcon Guardian; licensing and pricing not stated in the post.","operatingModelImplications":"Requires an owner for AI-agent policy decisions separate from traditional endpoint alerts.","sourceVerification":{"openedUrl":"https://www.crowdstrike.com/en-us/blog/crowdstrike-agent-graph-shows-how-agents-act-on-endpoint/","referenceExcerpt":"Agent hooks are available today for Claude Code, OpenAI Codex, GitHub Copilot, OpenCode, and Cursor.","promptVersion":"sled-research-v4.0","model":null,"basis":"agent-reported inspection"}}}]}