What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Pause the agent, contain the tools and systems it can reach, and invalidate the credential and any still-active sessions or delegated access. At the same time, preserve short-retention evidence. Then trace what the agent accessed or changed, remediate safely, and restore it only with credentials and permissions limited to its task.
Start by coordinating containment
Treat confirmed unauthorized access or credible exposure of a secret as a security incident. Suspicious model output by itself does not prove a credential was compromised. Record the detection time, affected agent or workload identity, suspected credential, known connected systems, and the incident lead. Follow your organization’s incident-response and communications plans. NIST SP 800-61 Rev. 3, published in April 2025, is the current revision identified here and supersedes Rev. 2; it integrates incident response with cybersecurity risk management. Read NIST SP 800-61 Rev. 3.
As an Amazon Associate I earn from qualifying purchases.
Distinguish a stolen or exposed credential from a legitimate credential that grants more access than the agent’s task needs. Both call for containment, but the investigation differs: a compromised secret may have been used by someone else, while excessive permissions may have enabled unintended actions without theft.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Pause the agent without losing sight of its dependencies
Use the platform’s controls to stop the agent’s current run and, as appropriate, disable scheduled tasks, queued work, and tool execution. Restrict workload or network access if that is part of the containment plan. Gate high-impact integrations while you determine which credentials and permissions they use. Coordinate disruptive changes with the incident lead; stopping the process alone does not invalidate tokens or end other access paths.
#1 Best Overall
- POWERFUL SECURITY KEY: The Security Key C NFC is the essential physical passkey for protecting your digital life from phishing attacks. It ensures only you can access your accounts.
- WORKS WITH 1000+ ACCOUNTS: Compatible with Google, Microsoft, and Apple. A single Security Key C NFC secures 100 of your favorite accounts, including email, password managers, and more.
- FAST & CONVENIENT LOGIN: Plug in your Security Key C NFC via USB-C and tap it, or tap it against your phone (NFC) to authenticate. No batteries, no internet connection, and no extra fees required.
- TRUSTED PASSKEY TECHNOLOGY: Uses the latest passkey standards (FIDO2/WebAuthn & FIDO U2F) but does not support One-Time Passwords. For complex needs, check out the YubiKey 5 Series.
- BUILT TO LAST: Made from tough, waterproof, and crush-resistant materials. Manufactured in Sweden and programmed in the USA with the highest security standards.
Containment and evidence preservation must proceed together. CISA advises preserving volatile or short-retention evidence, including system memory and firewall log buffers, before those records are lost. See CISA’s #StopRansomware Guide.
How do I revoke an AI agent’s credentials?
First identify the principal and credential type. The agent may use an API key, OAuth access or refresh token, workload identity, cloud role, service account, session cookie, signing key, or another authenticator. Trace who issued it, which services accept it, its lifetime, audience and scopes, and whether it was exchanged for or used to create other credentials.
- Revoke or disable the credential at its issuer. Use the issuer’s supported process for that credential type, and record when the action was requested and confirmed.
- Check every relying service and active access path. Determine whether existing access tokens, refresh grants, sessions, federated assertions, delegated permissions, API keys, or downstream credentials remain usable. Invalidate or terminate those paths where the provider and service support it.
- Look for credentials derived from the affected identity. Review whether the identity created, exchanged, or exposed other secrets, grants, keys, or sessions, and revoke affected access as appropriate.
- Verify the result. Check issuer and service audit events or other provider-supported status indicators; do not assume that a successful request at one system immediately invalidated access everywhere.
NIST IR 8587, finalized September 15, 2026, addresses token and assertion protection and lifecycle controls across identity providers, authorization servers, SSO, federation, APIs, and workload access. Its guidance is relevant when tracing how identity tokens move between an agent and connected services. See NIST IR 8587.
Free tools Windows power users keep installed
One-click scans. No signup required.
Rank #2
- POWERFUL SECURITY KEY: The YubiKey 5C NFC is the most versatile physical passkey, protecting your digital life from phishing attacks. It ensures only you can access your accounts
- WORKS WITH 1000+ ACCOUNTS: Compatible with popular accounts like Google, Microsoft, and Apple. A single YubiKey 5C NFC secures 100+ of your favorite accounts, including email, password managers, and more
- FAST & CONVENIENT LOGIN: Plug in your YubiKey 5C NFC via USB and tap it, or tap it against your phone (NFC), to authenticate. No batteries, no internet connection, and no extra fees required
- MOST SECURE PASSKEY: Supports FIDO2/WebAuthn, FIDO U2F, Yubico OTP, OATH-TOTP/HOTP, Smart card (PIV), and OpenPGP. That means it’s versatile, working almost anywhere you need it
- PRIMARY & SPARE KEYS: Just like having a spare house key, we recommend buying two YubiKeys - one for daily use and one as a spare. That way you’ll never get locked out of your accounts
For compromised authenticators within its scope, NIST SP 800-63B says: “The CSP SHALL suspend, invalidate, or destroy compromised authenticators from the subscriber’s account promptly following compromise detection.” It also says organizations SHOULD establish time limits for this process. This is not a universal API-token procedure: check the applicable authenticator requirements and the provider’s revocation behavior. See NIST SP 800-63B.
Do not equate stopping a process with revoking access
There is no single revocation command or propagation time established for every vendor and credential type. A stopped agent may have already obtained a token, established a session, or passed access to a connected service. Confirm what the issuer revokes, what each relying service accepts, and which sessions or grants need separate action.
Preserve evidence before retention windows close
Capture evidence that can establish what happened without copying secrets into tickets or ordinary logs. Record time zones and clock sources so identity, agent, cloud, application, and network events can be correlated. Limit access to incident records and preserve originals according to your organization’s evidence-handling procedures.
Rank #3
- POWERFUL SECURITY KEY: The YubiKey 5 NFC is the most versatile physical passkey, protecting your digital life from phishing attacks. It ensures only you can access your accounts
- WORKS WITH 1000+ ACCOUNTS: Compatible with popular accounts like Google, Microsoft, and Apple. A single YubiKey 5 NFC secures 100+ of your favorite accounts, including email, password managers, and more
- FAST & CONVENIENT LOGIN: Plug in your YubiKey 5 NFC via USB and tap it, or tap it against your phone (NFC), to authenticate. No batteries, no internet connection, and no extra fees required
- MOST SECURE PASSKEY: Supports FIDO2/WebAuthn, FIDO U2F, Yubico OTP, OATH-TOTP/HOTP, Smart card (PIV), and OpenPGP. That means it’s versatile, working almost anywhere you need it
- PRIMARY & SPARE KEYS: Just like having a spare house key, we recommend buying two YubiKeys - one for daily use and one as a spare. That way you’ll never get locked out of your accounts
- Agent identity, owner, run or task identifiers, and relevant prompts or triggering inputs.
- Tool-call parameters, results, policy decisions, approval records, and agent execution events.
- Identity-provider events, cloud and application audit trails, network records, and workload or container events.
- Volatile system state and short-retention records, such as memory and firewall log buffers, where available and authorized.
CISA highlights preservation of volatile and limited-retention evidence. OWASP recommends logging agent decisions, tool calls, and outcomes while warning against logging credentials or personally identifiable information in plain text. CISA’s guide and the OWASP AI Agent Security Cheat Sheet provide related guidance.
How can I tell what an AI agent did with a compromised token?
Build a timeline from the earliest plausible exposure through containment and effective revocation. Correlate events using the agent identity, principal, credential or session, source workload, tool, and target resource. Where records allow, establish which actions were performed by the agent and which may have come from another user or process using the same identity.
Scope both completed actions and reachable access. Determine what information was read, exported, altered, deleted, or transmitted; which resources were targeted; and whether the identity changed permissions, created credentials or persistence, or enabled other agents or systems to reuse its access. Review activity across connected services, not just the agent platform: a credential incident can extend to any service that shares, derives, or accepts the affected access.
Rank #4
- POWERFUL SECURITY KEY: The Security Key NFC is the essential physical passkey for protecting your digital life from phishing attacks. It ensures only you can access your accounts.
- WORKS WITH 1000+ ACCOUNTS: Compatible with Google, Microsoft, and Apple. A single Security Key NFC secures 100 of your favorite accounts, including email, password managers, and more.
- FAST & CONVENIENT LOGIN: Plug in your Security Key NFC via USB-A and tap it, or tap it against your phone (NFC) to authenticate. No batteries, no internet connection, and no extra fees required.
- TRUSTED PASSKEY TECHNOLOGY: Uses the latest passkey standards (FIDO2/WebAuthn & FIDO U2F) but does not support One-Time Passwords. For complex needs, check out the YubiKey 5 Series.
- BUILT TO LAST: Made from tough, waterproof, and crush-resistant materials. Manufactured in Sweden and programmed in the USA with the highest security standards.
Look for unusual behavior such as elevated privilege use, abnormal tool-call frequency, approval-bypass attempts, and sudden increases in high-risk actions. OWASP’s agent guidance recommends monitoring these patterns and keeping records of decisions, tool calls, and outcomes. OWASP AI Agent Security Cheat Sheet.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Remediate effects and restore the agent with less authority
Validate suspected changes with the owners of affected systems. Preserve the original state and coordinate any rollback; reversing an action can be unsafe, unauthorized, or impossible, and may create secondary effects. For destructive or high-impact operations, use recovery procedures that account for irreversible changes. OWASP Cornucopia recommends reversible transactions or dry-run modes where feasible and recovery exercises for agent-induced mass changes. See OWASP Cornucopia Agentic AI.
Before restarting, make reauthorization a deliberate decision rather than replacing the old secret with a similarly broad one. Remove access the task does not require, provision replacement credentials through approved secure channels, and test the agent on a limited task. Confirm that audit logging and alerting work before expanding its operation.
Best Value
- POWERFUL SECURITY KEY: The YubiKey 5 is a versatile physical passkey that protects your digital life from phishing attacks. It ensures only you can access your accounts.
- WORKS WITH 1000+ ACCOUNTS: Compatible with popular accounts like Google, Microsoft, and Apple. A single YubiKey 5 secures 100+ of your favorite accounts, including email, password managers, and more.
- FAST & CONVENIENT LOGIN: Plug in your YubiKey 5 via USB and tap it to authenticate. No batteries, no internet connection, and no extra fees required.
- MOST SECURE PASSKEY: Supports FIDO2/WebAuthn, FIDO U2F, Yubico OTP, OATH-TOTP/HOTP, Smart card (PIV), and OpenPGP. That means it’s versatile, working almost anywhere you need it.
- BUILT TO LAST: Made from tough, waterproof, and crush-resistant materials. Manufactured in Sweden and programmed in the USA with the highest security standards.
- Scope permissions to the specific tools, resources, actions, tenants, and environments the task needs.
- Separate credentials across integrated systems so one agent credential does not unnecessarily grant access everywhere.
- Use short-lived or otherwise lifecycle-managed access where supported, and define how active sessions and delegated grants are invalidated.
- Require independent policy checks or human approval for high-impact operations; fail closed if approval validation, policy lookup, or audit logging fails.
- Keep authorization, approval, execution result, and policy version attributable to the agent and task without recording secret material.
OWASP recommends minimum task-specific tool access, per-tool permission scoping, and separation of tools by trust level. It also recommends explicit authorization for sensitive operations and action previews or human approval for high-impact actions. OWASP AI Agent Security Cheat Sheet. The exact reactivation checks depend on the platform and your organization’s controls.
Use these criteria to assess whether controls are sufficient
Use the following questions to evaluate a design or post-incident change; they are practical criteria, not a published benchmark.
- Revocation reach: Can responders invalidate the agent key, active sessions, refresh grants, delegated access, and downstream credentials that remain usable?
- Scope: Can access be restricted by tool, resource, action, tenant, environment, and task?
- Time: Are credentials short-lived where appropriate, and can issuers and relying services invalidate them in a known, verifiable way?
- Attribution: Can records connect the owner, agent, credential, task, tool invocation, target, and outcome without exposing secrets?
- Containment impact: Can responders pause one agent or integration without unnecessarily disabling unrelated services?
- Recovery: Can high-impact operations be previewed, approved, reversed where safe, or reconstructed from protected records?
These criteria reflect NIST guidance on token lifecycle and incident response, alongside OWASP guidance on least privilege, auditability, and reversibility. NIST IR 8587, NIST SP 800-61 Rev. 3, OWASP AI Agent Security Cheat Sheet, and OWASP Cornucopia Agentic AI.
Recommended Free Tools
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
