Scan connected code repositories
Connect GitHub, GitLab, or Bitbucket. Inventory identifies supported AI applications, agents, and dependencies visible in the repositories you authorize.
Shadow AI discovery
See the AI your teams build, deploy, and use
Connect code repositories, cloud environments, AI infrastructure, and third-party tools to uncover AI outside your approved inventory. Add visibility into supported agents on employee workstations where Workstation Lens is deployed.
The challenge
AI can appear outside your approval process
An agent may start in a repository, use a cloud model, and connect to data or tools before security has a record of it. An asset list alone does not show those relationships or who owns the workflow.
The Repello approach
Inventory scans configured code, cloud, AI infrastructure, and third-party integrations for supported AI assets, then maps their relationships. Workstation Lens identifies supported agents and files on employee workstations where it is deployed. Coverage depends on the integrations configured and the workstations in your deployment.
How Repello helps
Map AI in cloud environments
Connect AWS, Azure, or GCP to bring supported cloud AI assets and their relationships into the same inventory.
Connect AI infrastructure and services
Extend discovery through supported Databricks, SageMaker, Kubernetes, OpenAI, Anthropic, and LangChain integrations.
See agents on employee workstations
Workstation Lens identifies supported coding agents, skills, MCP servers, and configuration files on workstations where it is deployed.
What Repello gives your team
Go from an asset list to a security decision
See what each application connects to, who owns it when available, and which exposures deserve review.
Understand each asset's reach
- Trace supported applications to the models, data sources, tools, and MCP connections they use.
- See available ownership and dependency context before assigning a review.
- Investigate supported on-device agents and the files connected to them.
Decide what to secure next
- Focus first on sensitive data access, broad tool permissions, and unknown ownership.
- Send an application for targeted ARTEMIS testing or an ARGUS policy review.
- Use Workstation Lens to investigate risky agent files on affected employee workstations.
Frequently asked questions
What to know about AI discovery.
How does Repello discover shadow AI?
Inventory scans the code, cloud, AI infrastructure, and third-party sources you connect for supported AI assets. It maps applications, agents, models, data sources, and tools where those integrations provide the information. Workstation Lens separately identifies supported agents and related files on employee workstations where it is deployed.
Which systems can Inventory connect to?
The Inventory product page lists GitHub, GitLab, and Bitbucket for code; AWS, Azure, and GCP for cloud; Databricks, SageMaker, and Kubernetes for AI infrastructure; and OpenAI, Anthropic, and LangChain under third-party tools. Confirm the specific integration and access available for your deployment.
What can we see after an AI asset is discovered?
Inventory maps supported relationships between an application and its models, data sources, tools, and MCP connections. Available owner information adds context for assigning a review. The detail shown depends on what each configured integration exposes.
Can Repello discover AI agents on employee workstations?
Yes. Workstation Lens identifies supported agents, skills, MCP servers, and configuration files on employee workstations where it is deployed. Coverage depends on the workstations and agent types included in your deployment.
Does Repello discover AI on every employee workstation or website?
No. Workstation Lens covers workstations where it is deployed, while Inventory sees assets available through configured integrations. Neither provides a complete record of activity on other workstations or every AI website employees visit.
Does discovering an unreviewed AI asset mean it is unsafe?
No. Discovery shows that the asset exists and, where available, what it connects to. Review its purpose, owner, data access, and tool permissions before deciding whether it needs a security test or control.
What should we do after discovering shadow AI?
Start with assets that reach sensitive data or powerful tools, or lack a clear owner. Bring the responsible team into the review, test relevant application behavior with ARTEMIS, consider an ARGUS policy for connected workflows, or investigate agent files with Workstation Lens.
See Repello in action
See your AI footprint with Repello.
Show us the sources you want to connect. See the applications, agents, models, data, and tools Repello can map, then decide what needs attention first.