ripgrep
Binary · BurntSushi
96 score
EDR Stops at the OS, Agents Operate Beyond It
Software Has Changed Get Control Over It
See, attribute, and control the agents and tools operating inside trusted software with one lightweight sensor.
Neoverse
The trusted catalog for AI software.
Neoverse catalog intelligence
Find Everything and Control It
Neoverse is our autonomous software knowledge base that constantly uncovers new capabilities and risks in agentic and non-agentic software. It powers our platform and is included with every deployment for no additional cost.
Neoverse identifies
- AI Agents
- MCP Servers
- Skills
- Models
- Binaries
1.2M+
agentic artifacts cataloged and risk-profiled
Continuously mapped across agents, skills, binaries, MCP packages, and third-party models.
Desktop AI Helper
by brightstack
Windows assistant with broad local privileges and cross-app automation.
Operator Scout
by northstar-ai
Autonomous browser operator for research and repetitive web workflows.
Workflow Pilot
by relayworks
Coordinates application tasks with delegated tool and data access.
Representative catalog results · Modeled on the Neoverse product experience
notion-sync-mcp
by nclabs
Two-way sync between Notion workspaces and local knowledge stores.
postgres-mcp
by crystaldata
Query and inspect PostgreSQL databases through an agent toolset.
slack-mcp
by modelcontextprotocol
Read messages, list channels, and search workspace conversations.
Representative catalog results · Modeled on the Neoverse product experience
invoice-helper
by finflow-tools
Extracts invoice line items and prepares data for finance systems.
repo-reviewer
by shipshape-ai
Reviews pull requests and proposes changes across repositories.
browser-research
by fieldnotes
Collects web sources and synthesizes evidence for research tasks.
Representative catalog results · Modeled on the Neoverse product experience
Reasoning Model X
by external-inference
Long-context reasoning model connected to developer workflows.
Code Model Pro
by vectorforge
Code generation model with repository and terminal context access.
Vector Embed 4
by openvector
Embedding model used to index internal document collections.
Representative catalog results · Modeled on the Neoverse product experience
claude-cli-helper
by Unverified
CLI package distributed under a name similar to a trusted tool.
agent-runtime
by runlayer
Local runtime that brokers model access and tool execution.
desktop-bridge
by commonware
Native bridge for desktop agents and local applications.
Representative catalog results · Modeled on the Neoverse product experience
The Gap in the Current Stack
Trusted Applications Can't Be Trusted
Endpoint Control guards the operating system. Neo governs the human and non-human behavior happening above that line.
Current controls see
Trusted application
Examples include Claude, GitHub Copilot, OpenAI, Google Gemini, Perplexity, Mistral.
- 01 Actors multiply
Autonomous Actors Appear Inside the App
Agents, plugins, MCP servers, and embedded models can change without the application binary changing.
- 02 Identity blurs
One Valid Session Masks Who Acted
A person, an agent, and its connected tools can all operate through the same trusted identity.
- 03 Intent changes
Approved Actions Might Be Malicious
Credential access and data movement can look legitimate to controls that only see the sanctioned app.
The application still looks trusted from the outside. The security-relevant behavior has moved inside it.
For the evaluation team
Validate the Control, Not Just the Claim.
Start with the questions that determine whether a platform can govern agentic behavior in production.
- < 15 min
- First scan to full inventory
- 25MB
- Lightweight endpoint agent with full control
- < 1 Hour
- Full deployment with IdP and SOC
How does Neo fit with the controls we already own?
Endpoint, identity, network, and data controls continue to protect their existing boundaries. Neo adds visibility, attribution, and enforcement inside the application layer those tools treat as trusted.
What does Neo see that endpoint inventory does not?
Neo identifies the agents, plugins, MCP servers, skills, extensions, and embedded AI functions operating inside an application—not only the process or binary that contains them.
What does runtime control mean?
Neo evaluates an action while it is happening and can allow, block, or hold it before data moves. The decision is preserved with the actor, action, and policy context.
Can Neo distinguish a person from an agent using their session?
Yes. Neo ties application activity to both the human session and the non-human actor that initiated it, preserving the attribution traditional application logs lose.
What will we see in a demo?
We will walk through discovery, attribution, and a runtime policy decision so your security and technical teams can evaluate the complete control loop together.
Get a Demo
See the Agentic Application Layer in One Control Loop
Bring your security and technical evaluators. We will show how Neo discovers the actor, attributes the action, and enforces policy before data moves.