Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Contain first, investigate second. Freeze the suspected agent or workflow, block its principal, and revoke every active credential it could use. Preserve prompts, tool calls, environment snapshots, and audit events before destructive cleanup. Then inventory the full trust surface, issue least-privileged replacements through a controlled issuer, deploy and test them, retire the exposed credentials, and investigate activity from the earliest possible exposure time.
NIST’s rule is unambiguous: “The CSP SHALL suspend, invalidate, or destroy compromised authenticators from the subscriber’s account promptly following compromise detection.”
As an Amazon Associate I earn from qualifying purchases.
The first response: stop the agent’s access
Treat a suspected compromise as an active incident, even if you do not yet know whether an attacker used the exposed secret. An agent can pass credentials to tools, workflows, build systems, repositories, cloud APIs, and model providers. Waiting for a complete explanation gives an intruder more time to use the same identity or establish persistence.
PC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minuteFreeze the execution path
Pause the agent service, tool connector, scheduled workflow, runner, or deployment associated with the alert. Stop new jobs and prevent automatic retries. If a shared platform cannot pause only one run, suspend the affected project or workload long enough to block further calls. Record the time of the decision and the last known execution so the investigation has a defensible starting point.
#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.
Block the principal without destroying evidence
Disable the agent’s service account or apply an IAM deny policy that blocks its principal. A deny policy is preferable when the platform supports it and you need to preserve the identity, role bindings, and audit history for investigation. Do not delete the principal before exporting the records that identify its permissions, token issuances, and recent activity.
Revoke every credential in the affected path
At the containment gate, revoke or disable API keys, OAuth access and refresh tokens, GitHub App credentials, OIDC trust grants, cloud service-account keys, package-registry tokens, signing keys, and model-provider keys associated with the agent or the suspect run. Include credentials used by the same runner, connector, deployment job, or environment; an agent may have read a secret without using it immediately.
Preserve volatile evidence
Before deleting files or rebuilding hosts, export prompts, model responses, tool-call records, shell history, environment snapshots, workflow logs, container and image metadata, and provider audit events. Capture timestamps in a consistent time zone and record which credentials were disabled and when. This evidence supports scope analysis and helps distinguish an accidental leak from deliberate use.
OWASP’s incident-response guidance states: “Contain — revoke or rotate the credential immediately.” It also calls for querying audit logs for all activity by that credential since exposure.
Rank #2
Yubico - YubiKey 5C NFC - Multi-Factor authentication (MFA) Security Key and passkey, Connect via USB-C or NFC, FIDO Certified - Protect Your Online Accounts
- 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
Map the complete credential and trust surface
An agent’s environment is larger than its visible API key. Build an inventory before issuing replacements, and keep the compromised entries in it until revocation and verification are complete. For each item, record the credential identifier, issuer, privilege, environments, storage locations, last-seen use, expiration, replacement owner, and revocation method.
| Surface | What to check | Containment and follow-up |
|---|---|---|
| Agent runtime | Environment variables, mounted files, .env files, command arguments, shell history, prompts, traces, and logs |
Revoke exposed values; remove copies from state, traces, images, and caches |
| CI/CD and build systems | Pipeline variables, workflow artifacts, caches, build logs, deployment manifests, runner tokens, and container layers | Pause affected jobs, rotate variables and runner credentials, and inspect artifacts produced during exposure |
| Cloud IAM | Service accounts, access keys, role bindings, workload-identity federation, token-creator permissions, and newly created principals | Disable or deny the principal, rotate keys, review grants, and investigate new identities |
| Source-control integrations | App installations, OAuth grants, deploy keys, webhook secrets, and personal-access-token substitutes | Remove unauthorized installations or grants and rotate repository and webhook secrets |
| Registries and signing | Package-registry tokens, image-push credentials, code-signing keys, provenance keys, and publication permissions | Revoke tokens and signing authority; quarantine packages or images created in the exposure window |
| Model and tool providers | Model-provider keys, tool-connector tokens, browser or database credentials, and delegated OAuth scopes | Revoke at the provider, reduce scopes on replacements, and review usage and billing records |
| Human and downstream accounts | Shared operator credentials, notification accounts, ticketing integrations, and secrets passed into prompts | Rotate any value the workflow could read, even when the agent itself did not authenticate with it |
Search both live systems and historical material. Secrets can remain in source repositories, old container layers, crash dumps, workflow artifacts, log archives, prompt transcripts, and local developer caches. A secret that was merely displayed to a model or written into a trace must be treated as exposed.
Choose a revocation and replacement strategy
Revocation speed, maximum token lifetime, privilege scope, auditability, blast radius, recovery path, provider support, and operational effort are the practical comparison axes. The right design depends on what the issuer can invalidate and how quickly every consumer can receive a replacement.
| Credential or grant | Immediate action | Replacement design questions |
|---|---|---|
| Static API or model-provider key | Disable or revoke at the issuing service | Can the issuer create a narrower key, set an expiration, attribute calls to one workload, and show complete usage logs? |
| OAuth access or refresh token | Revoke the refresh token and active access tokens where supported; remove unauthorized grants | What is the provider’s revocation latency, token lifetime, scope model, and re-consent process? |
| OIDC or workload-identity trust | Remove or narrow the trust relationship and deny the affected principal | Can tokens be issued only for a specific workload, audience, repository, or environment with automatic expiry? |
| Cloud service-account key | Disable the old key, verify the new one, then delete the old key | Can the workload use federation instead of a static key, and who owns role and token-creator permissions? |
| Package, deployment, or signing credential | Revoke publication or signing authority and quarantine outputs | How will you re-establish provenance, invalidate affected artifacts, and distribute a trusted replacement? |
| Secrets embedded in prompts, logs, or images | Assume disclosure and rotate at the original issuer | Where else was the value copied, and can retention, redaction, and access controls be tightened? |
Do not confuse disabling an identity with rotating a secret. A disabled principal may still have valid tokens or keys accepted elsewhere, while a rotated key does not stop an attacker who can obtain a second credential through the same trust path. Use both controls when the platform allows it.
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
Rotate credentials through a controlled issuer
Perform replacement in a sequence that keeps the service recoverable and prevents the old value from returning through automation.
- Define the required access. List the exact APIs, repositories, resources, and actions the workload needs. Separate read-only operations from writes and remove administrative privileges.
- Create a replacement. Issue a new credential with the smallest practical scope and shortest practical lifetime. Prefer OIDC or workload identity and a secrets manager over a static key. Give the replacement a distinct identifier so audit logs can separate old and new activity.
- Update the source of truth. Store the replacement only in the approved secrets manager or identity configuration. Remove hard-coded values from code, manifests, prompts, images, and local configuration. Check that templates and deployment automation cannot reintroduce the exposed value.
- Deploy everywhere the old value was used. Update every service, job, region, connector, and failover environment. Include dormant schedules and emergency paths; an overlooked worker can continue using the compromised credential.
- Verify with the new identity. Test authentication, authorization, health checks, and the specific read or write operations the workload requires. Confirm that denied operations remain denied and that rollback instructions are ready.
- Disable the old credential and observe. Watch service errors, queue retries, and dependency alerts after disabling it. A failed call identifies a consumer that was missed during deployment; do not re-enable the exposed value merely to make the error disappear.
- Delete or permanently revoke the old value. Remove it from issuer inventories and then purge copies from code history where possible, logs, images, caches, artifacts, and agent state. Keep the incident record and forensic evidence, not the usable secret.
- Confirm expiry and residual access. Check that outstanding sessions or tokens have expired or were explicitly invalidated. Review issuer logs for attempts to use the old identifier after cutover.
Google Cloud service-account key sequence
For a Google Cloud service account, the documented recovery sequence is to create and deploy a new key, disable the old key to verify that the replacement works, delete the old key, revoke token-creator permissions if abuse is possible, and wait for token expiry before re-enabling a disabled service account. Prefer workload-identity federation for the long-term design so a file-based key is not required.
Investigate what the agent accessed
Query provider and platform audit logs from the earliest possible exposure time, not merely from the moment someone noticed the incident. Preserve the original records in tamper-evident storage and export them before retention windows close.
Recommended Free Tools
- Authentication events, token issuance, refresh, failed logins, and calls from unusual IP addresses, regions, runners, or user agents.
- API reads, exports, downloads, database queries, object access, and messages sent to external destinations.
- Privilege changes, role-binding edits, policy changes, newly created keys, new identities, OAuth grants, app installations, and webhook modifications.
- Resource creation, modified workflows, scheduled jobs, repositories, branches, packages, images, and deployment targets.
- Attempts to disable logging, alter retention, delete evidence, or create persistence through automation.
Correlate records by credential identifier, principal, workload, source address, and timestamp. A clean application log does not prove that no access occurred: the issuer’s audit trail may contain calls made through a delegated token or a connector.
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.
Trace prompts, tools, and downstream artifacts
Review the agent’s prompts and responses for secrets it was shown, instructions that changed its behavior, and tool calls that created or modified resources. Use build provenance or an AI bill of materials to identify packages, images, pull requests, configuration files, and deployments produced under the suspect identity. Quarantine those artifacts until they are reviewed.
Eradicate persistence and recover from a known-good state
- Remove secret material. Delete exposed values from source, images, logs, traces, caches, notebooks, and agent memory or state stores. Rotation is still required because deletion cannot prove that a copy was not made.
- Rebuild cleanly. Recreate affected images, packages, workflows, and infrastructure from a known-good commit or image. Do not “repair” an artifact whose build inputs or provenance cannot be trusted.
- Redeploy with the replacement identity. Confirm the new scopes, audience, environment bindings, and expiration settings in each deployment.
- Validate dependencies. Run functional and security checks, inspect outbound connections, and verify that monitoring and audit export are active before restoring normal traffic.
- Notify affected parties. Inform downstream consumers, repository owners, cloud administrators, or customers when their data, artifacts, or trust relationships may have been affected. Follow contractual and legal notification requirements for your jurisdiction.
- Close the incident with evidence. Record the exposure window, credentials revoked, systems checked, artifacts quarantined, notifications sent, and remaining uncertainty.
Design the next deployment to limit blast radius
Give every agent an attributable identity
Create a dedicated identity for each agent or narrowly defined workload. Never place a developer’s personal credential in an agent environment, and do not reuse one production identity across unrelated agents. Separate read-only and write-capable identities so a prompt injection cannot turn a data-reading task into an administrative action.
Use short-lived, task-scoped access
Issue credentials for a specific task, audience, repository, or environment and let them expire automatically. OIDC and workload identity reduce the number of static secrets that can be copied. When a static credential is unavoidable, keep its scope narrow, set the shortest practical lifetime, and document the issuer’s emergency revocation path.
Make tools deny-by-default
Allow only the tools and operations required for the current task. Require human approval for destructive, external, financial, privilege-changing, or production actions. Treat tool arguments, retrieved documents, and model output as untrusted input; a model instruction must not be able to grant itself a new permission.
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.
Centralize secrets and make use observable
Use a secrets manager or workload-identity service rather than environment files or prompts. Alert on anomalous token use, new principals, unusual regions, bulk reads, publication events, and changes to logging or trust policy. Store audit records in tamper-evident storage with retention long enough to cover the maximum investigation window.
Exercise the playbook
OWASP’s AI verification guidance calls for a written playbook covering agent-credential revocation, rotation of every secret touched by a compromised workflow, artifact quarantine, downstream notification, preservation of prompts and audit logs, and testing of a measured time-to-revoke. Run at least annual tabletop or live-fire exercises for prompt injection, suspicious pull requests, stolen runner tokens, and malicious workflow changes. Measure the time from detection to blocked access, replacement deployment, and confirmed deletion of the old credential.
Use a physical authenticator as a recovery option
NIST identifies a physical authenticator as an acceptable backup or alternate authenticator after loss, theft, or compromise. A hardware security key can therefore be part of the recovery plan for administrators or operators who must re-enroll accounts after an incident. It does not replace revoking an agent’s machine credentials; it protects the human account used to perform recovery.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Before purchasing or standardizing one, confirm protocol support, phishing resistance, account-assurance compatibility, enrollment requirements, durability, centralized enrollment and revocation, and the organization’s backup-key policy. Keep a separately protected backup key and document who can authorize enrollment or revocation. Availability and behavior vary by identity provider, so test the complete recovery path with the actual accounts and policies in use.
Common mistakes that prolong exposure
| Mistake | Why it fails | Better response |
|---|---|---|
| Waiting for forensic certainty | The attacker retains a usable token while the team investigates | Block the principal and revoke credentials immediately, then investigate with preserved evidence |
| Rotating only the visible API key | Refresh tokens, app grants, cloud keys, or copied secrets remain valid | Inventory every issuer and trust relationship touched by the workflow |
| Deleting the identity first | Permission history and attribution can be lost | Export logs and snapshots, then disable or deny while preserving the principal where possible |
| Deploying a replacement without testing | Hidden consumers fail or teams restore the exposed key | Verify each service, region, connector, and rollback path before retirement |
| Leaving leaked values in logs or images | Anyone with historical access can recover the secret after rotation | Purge copies, restrict historical access, and rotate again if a copy remains usable |
| Reusing a broad production identity | One compromise creates a large, hard-to-audit blast radius | Use dedicated, least-privileged, short-lived identities with attributable activity |
Standards and guidance to keep with the runbook
NIST’s Digital Identity Guidelines, SP 800-63B, requires prompt suspension, invalidation, or destruction of compromised authenticators. NIST published SP 800-61 Rev. 2, the Computer Security Incident Handling Guide, in 2012 and issued Rev. 3 community-profile guidance in 2025. OWASP’s DevSecOps incident-response material emphasizes immediate credential revocation or rotation and audit-log review; its AI-agent guidance emphasizes dedicated identities, scoped short-lived tokens, and avoiding long-lived production credentials in environments, prompts, or configuration files.
No authoritative source establishes a universal breach rate or a universal time-to-revoke target. Set an internal target based on the providers you use, test it during exercises, and improve the measured result after every incident.
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.
