Skip to content
The State of Exposed AgentsRead the report
Guard0

Reads what you run. Reports where you work.

Every connection on the way in is read-only. On the way out, each finding arrives with its evidence in the tools your team already watches.

Reads from

Read-only, every one.

  • Code hosts

    Repositories, where agents, tools, prompts and MCP servers are defined.

    • GitHub
    • GitLab
    • Bitbucket
    • Azure DevOps
  • Clouds

    Cloud accounts, through read-only roles: the models, endpoints and agents deployed there.

    • AWS
    • Azure
    • Google Cloud
    • Oracle Cloud
  • Model platforms

    The model providers your agents call.

    • OpenAI
    • Anthropic
    • Google Gemini
    • Vertex AI
    • Azure OpenAI
    • Amazon Bedrock
    • Mistral
    • Cohere
    • Hugging Face
    • Ollama
  • Agent platforms

    The agents inside the platforms you buy, through their admin and audit APIs.

    • Copilot Studio
    • Microsoft 365 Copilot
    • Agentforce
    • ServiceNow
    • Glean
    • ChatGPT Enterprise
  • Agent frameworks

    The frameworks your teams build agents with, found in their code.

    • LangChain
    • LangGraph
    • CrewAI
    • AutoGen
    • OpenAI Agents SDK
    • Bedrock AgentCore
    • LlamaIndex
    • Semantic Kernel
    • Vercel AI SDK
    • n8n
    • Zapier
    • Make
    • Dify
    • Flowise
    • Langflow
  • Coding agents

    The coding agents on your developers’ laptops.

    • Claude Code
    • Cursor
    • GitHub Copilot
    • Windsurf
    • Kiro
    • Amp
    • Gemini CLI
    • Amazon Q
    • Cline
  • MCP servers

    MCP servers, in repositories and on machines.

    • Any MCP server
    • OpenClaw
  • Identity

    Who each agent acts as, and who owns it.

    • Okta
    • Microsoft Entra ID
    • Google Workspace
    • Ping Identity
    • OneLogin
    • JumpCloud
    • Auth0
  • Laptops, through MDM

    Laptops, through a read-only sensor your MDM deploys.

    • Jamf
    • Intune
    • Iru
  • Secrets

    Where agents’ credentials are kept, and which agent holds which.

    • HashiCorp Vault
    • AWS Secrets Manager
    • CyberArk
    • 1Password
    • Doppler
  • Data

    The data your agents can reach.

    • Snowflake
    • Databricks
    • BigQuery
    • PostgreSQL
    • Salesforce
    • Dynamics 365

And many more, all read-only.

Reports to

Each finding, with its evidence.

  • Chat

    Findings as messages, to the people who need to see them.

    • Slack
    • Microsoft Teams
    • Email
  • Tickets and on-call

    Findings as tickets and pages, each with an owner.

    • Jira
    • ServiceNow
    • Linear
    • PagerDuty
    • Opsgenie
  • SIEM and observability

    Findings and their evidence, as events.

    • Splunk
    • Microsoft Sentinel
    • Google SecOps
    • Datadog
    • Elastic
    • Sumo Logic
    • Panther
    • Exabeam
  • EDR and XDR

    Findings shared with the tools that watch your devices.

    • CrowdStrike
    • SentinelOne
    • Microsoft Defender
    • Cortex XDR
    • Trend Micro
  • Network and SSE

    Findings shared with the tools that watch your traffic.

    • Zscaler
    • Netskope
    • Prisma Access
    • Cloudflare
    • Fortinet
    • Cisco
  • Cloud and app security

    Findings beside your cloud and application risk.

    • Wiz
    • Orca
    • Prisma Cloud
    • FortiCNAPP
    • Tenable
    • Qualys
    • Rapid7
    • Snyk
  • GRC

    Evidence for your audits and frameworks.

    • Vanta
    • Drata
    • OneTrust
    • Archer
  • CI gates

    Checks that stop an unsafe agent before it ships.

    • GitHub Actions
    • GitLab CI
    • Jenkins
    • CircleCI

And anything else with a webhook.

Missing a tool your team relies on? Tell us about it.

Start free.

Find the agents already running in your company, and who answers for each one.

Free for up to 3 people. No card. Team from $416 a month, billed annually. See pricing