This website counts page views with Cloudflare Web Analytics, which sets no cookies and stores nothing on your device. If you choose Allow, embedded content such as Spotify podcast players also sets optional cookies. Choose Deny to turn off both. Cookie Notice
An agent can use the accounts its user is signed in to, and nobody has a list of which agent can reach which identity.
Agents are set up with accounts for GitHub, cloud providers, and business apps. They also inherit what the environment they run in can reach: AWS and Google Cloud profiles, Kubernetes contexts, and SSH keys loaded for the session. Identity tools see the account, and device tools see the app. Neither shows which agent can use which identity.
Research from across the industry shows how much access already sits where agents can reach it.
24,008
unique secrets exposed in MCP-related configuration files on public GitHub.
2,117 of them were valid credentials, from GitGuardian's scan of public GitHub in 2025.
The files that connect agents to tools often hold live credentials.
GitGuardian, State of Secrets Sprawl 2026, March 2026
2,000+
unique, verified secrets leaked in the s1ngularity attack, which told local AI command line tools to search for credentials.
Over 1,700 users had secrets leaked; about half of all victims had at least one AI command line tool installed.
Whatever identities an agent can reach, an attacker can reach through it.
A map of every agent and the identities it can reach.
Identity map
Which agent can use which identity
Control Plane lists each agent on each device with the identities it's configured to use or may inherit, such as GitHub accounts, cloud profiles, Kubernetes contexts, and SSH keys, and how each one is reached.
Evidence
Clear about what it knows
Each link says whether the agent is configured with an identity, references a credential source, or has local evidence that the source exists, so nobody reads more into it than the device shows.
Findings
Spot the risky setups
Supply Chain flags credentials written into configuration files, a person's own account used by an agent, shared profiles, and cloud access an agent inherits without anyone choosing it.
Privacy
A full map with no secrets collected
Supply Chain builds the map from account names and non-secret metadata only. Passwords, tokens, private keys, browser sessions, and vault contents are never read.
How it works
Map access with the package you already run.
01 · Supply ChainDeploy once
Push one package through your device management tool, such as Jamf, Iru, or Intune. Identity discovery runs in the same scan as the agent inventory.
02 · Supply ChainReview the map
Select a device or agent in Control Plane and see each identity it can reach, how, and how strong the evidence is.
03 · Runtime, optionalAct on it
Move agents to their own managed identities where Control Plane recommends it, and add Runtime to check what agents do with the access they have.
Supply Chain / Agent accessIllustrative data
CodexEngineering · 1 device · identities
Configured · OAuth session
Human account · identity-7f3a
May inherit · shared user environment
Provider profile · references a credential source
Local evidence · SSH agent
Public key metadata only
Agent uses a person's own account
Recommended: a managed identity for the agent
In Control Plane
Each agent shows what it can reach.
Open any agent to see its identities, written plainly.
"Codex is configured with this GitHub identity through an OAuth session."
"Codex may inherit this AWS profile through its shared user environment."
Totals across the fleet by provider, kind of identity, and risk
What it costs
Priced per device. Always predictable.
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.
Your secrets stay where they are. Supply Chain maps access from account names, profile names, and other non-secret metadata, and never reads passwords, tokens, private keys, browser sessions, or password manager vaults.
Who can see the exact account names?
Only the people who need them: workspace owners and administrators. Everyone else sees a stand-in name for each identity, and every reveal of an exact name is recorded.
Which identity sources does it check?
Agent and connector configuration, plus developer identity sources such as GitHub CLI accounts, AWS, Google Cloud, and Azure profiles, Kubernetes contexts, and SSH agent keys.
Is this part of Supply Chain?
Yes. Identity discovery is part of Supply Chain and runs in the same package and scan as the agent inventory.
Modules used
Built on these modules.
Before trust
Supply Chain / InventoryIllustrative data
Supply Chaindevice:endpoint_4c19 trust:any⌘↵
3 agents14 components
Cursor
New agent found · Unreviewed
deploy-helper / v2.4.1
Critical finding · Blocked · Applied
ticket-tracker
Identified · Trusted
Supply Chain
Find every AI agent on every device. Check every skill it loads.
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.
Reduce agent risk without slowing your teams down.
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.