Agent Authenticator: Passkey-Grade, Hardware-Bound Authentication for AI Agents
Organizations: Independent Researcher
Abstract
AI coding agents now push code, publish packages, call model APIs and operate cloud resources on behalf of developers. They authenticate to these services with the same bearer secrets humans use, personal access tokens, API keys and package-registry tokens, stored where the model, its subprocesses and any co-resident malware can read them. Incidents in 2025 and 2026, including prompt-injection attacks on coding agents and self-propagating supply-chain worms, show that this arrangement fails in predictable ways. Passkeys solved the analogous problem for humans, but they require a human gesture on every use and cannot be operated by an autonomous process. We present the Agent Authenticator, a design that gives agents passkey-grade authentication properties: per-service hardware-bound keys, no shared secrets, origin binding enforced by the runtime rather than the model, normalized attestation of the execution environment, human-signed scoped mandates, and cross-device human approval for high-risk actions. The same protocol serves agents on developer laptops and in cloud enclaves, and it deploys as a profile over existing mechanisms, MCP authorization, OAuth token exchange including the Identity Assertion JWT Authorization Grant, git and SSH, rather than as a new identity system. The authenticator also produces a signed, hash-chained record of each action it authorizes. We analyze the design against a threat model centered on prompt injection and same-host compromise, map it onto five documented incidents, and discuss its limits.
Figures & tables
| Incident | What happened | Root cause |
|---|---|---|
| GitHub MCP prompt injection (2025) [ 12 ] | A malicious issue in a public repository caused an agent to read private repositories and publish their contents in a public pull request. | One broadly scoped token covered every repository the user could reach. |
| Comment and Control (2026) [ 13 ] | A crafted pull-request title hijacked several CI-hosted coding agents, which leaked repository secrets back through the code host itself. | Secrets were ambient in the agent’s environment; untrusted text was treated as instructions. |
| Coding-agent exploits (2025–2026) [ 14 ] | Crafted branch names exfiltrated tokens; repository settings files bypassed trust prompts; write sandboxes were escaped. | Controls ran inside the agent and could be disabled by repository content. |
| Nx “s1ngularity” (2025) [ 15 ] | Malicious package versions harvested 2,349 credentials from 1,079 developer machines, invoking locally installed AI CLIs to search for secrets. | Tokens and keys on disk were readable by any local process. |
| Shai-Hulud worm (2025–2026) [ 16 ] | A self-propagating worm stole npm and GitHub tokens and republished itself into packages the victims maintained. | Long-lived publish tokens were usable from any machine. |
| Key | Holder | Lifetime |
|---|---|---|
| Approver key | Human’s passkey authenticator | Long-lived; device-bound passkeys recommended |
| AA identity key | AA secure hardware | Life of the device or cloud environment |
| Agent credential key | AA secure hardware, one per agent and relying party | Local: until mandate expiry or revocation. Cloud: one task |
| Evidence key | AA secure hardware, separate from credential keys | Rotated per epoch, forward-secure |
| Binding | Mechanism | Role of the AA |
|---|---|---|
| MCP | MCP authorization with DPoP and workload-identity extensions [ 8 ] | Acts as the MCP client’s auth provider; tokens bound to the agent key |
| Enterprise IdP | ID-JAG: token exchange then JWT grant [ 7 , 3 , 4 ] | Agent assertion is the actor token; IdP applies policy |
| OAuth APIs | JWT grant or token exchange with RAR | Assertion redeemed for a key-bound access token |
| Git over HTTPS | git credential helper | Per-remote tokens; no tokens on disk |
| Git over SSH | ssh-agent with a key type modelled on FIDO SSH keys | Signatures carry flags and mandate hash |
| Cloud | Workload identity federation | Short-lived cloud credentials; no static keys |
| Incident | Outcome | Mechanism |
|---|---|---|
| GitHub MCP injection | Contained | Mandate limits the agent to named repositories; cross-repository publication outside them is refused. |
| Comment and Control | Blocked | No secrets exist in the agent’s environment; review agents receive no secret-reading scope. |
| Coding-agent exploits | Contained | Policy is enforced by the AA outside the agent, so repository content cannot disable it; tokens are not exposed to exfiltrate. |
| Nx s1ngularity | Blocked | No credentials on disk to harvest; hardware keys cannot be read. |
| Shai-Hulud worm | Blocked | Publishing requires per-action human approval, and credentials cannot be used from other machines. |
AI Agent Security
Artificial intelligence agent security studies attacks and safeguards for tool-using, persistent, and collaborative systems across digital and physical settings.
See all 558 papers