Free tools Windows power users keep installed
One-click scans. No signup required.
Revoke a compromised AI agent at the identity and authorization layers—not only by stopping its process. Pause the runtime, preserve logs, disable or deny its workload identity, revoke every OAuth and tool grant, rotate persistent secrets, and investigate access before rebuilding. Deleting one API key can leave short-lived tokens issued from the same service identity usable.
This runbook covers agent orchestrators, service accounts, API keys, OAuth/OIDC connections, tool or MCP servers, CI/CD systems, repositories, and downstream data stores. It is designed for an active incident as well as the hardening work that follows.
As an Amazon Associate I earn from qualifying purchases.
What to do in the first 15 minutes
Assume the agent is an identity incident. A model process may have several independent ways to authenticate, and a single agent can reach tools, repositories, cloud resources, and data stores through credentials issued by different providers.
Crashes, 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 minutePC 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 & 11- Declare and scope the incident. Record the affected agent, runtime or orchestration instance, owner, workload or service identity, API keys, OAuth and OIDC grants, tool-server connections, network paths, repositories, data stores, CI/CD jobs, and any other agents that share credentials.
- Preserve evidence. Export authentication, service-account, tool-call, network-egress, repository, and data-access logs before changing systems. Keep a controlled forensic copy of the runtime where practical. Google Cloud guidance recommends reviewing unfamiliar agent runtimes, sessions, service accounts, and identities and backing up logs for analysis. The Australian Cyber and Infrastructure Security Centre similarly emphasizes an inventory of AI accounts, service identities, and credentials and preserving logs before restoration.
- Stop new actions. Pause the workflow, disable its scheduler or webhook, quarantine the worker, and block egress to sensitive tools and data services. Do not destroy the only copy of the runtime or its logs while doing this.
- Cut off the principal at its issuer. Disable or delete the underlying service account, or apply an issuer-side deny control that blocks the principal. This is the step that can stop active tokens; deleting a key alone may not.
- Revoke delegated access. Remove OAuth grants, SSO authorizations, personal access tokens, deploy keys, SSH keys, and tool-server credentials at every connected SaaS, source-control, cloud, and data provider.
- Rotate and recover. Create replacement secrets only after the compromised path is blocked, deploy them to legitimate consumers, test them, and then disable and delete the old credentials. Investigate use before rebuilding the agent.
Keep a timestamped incident log showing who performed each action, which issuer or system was changed, and what evidence was retained. That record prevents a “key deleted” event from being mistaken for complete containment.
#1 Best Overall
- 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 every identity the agent can use
Build the map from configuration, deployment manifests, secret stores, runtime environment variables, identity-provider records, and provider audit logs. The agent’s visible API key is only one possible credential.
Runtime and orchestration identities
Record the worker, container, virtual machine, serverless function, scheduler, webhook identity, and orchestration account. Include temporary execution roles and any role that can impersonate another identity or mint tokens. Check whether another agent, job, or human process shares the same principal.
Cloud and service accounts
For each service account, capture its issuer, account identifier, roles, token-minting or impersonation permissions, key IDs, token audiences, expiration behavior, and last-use time. A service account that can create credentials for another account expands the incident beyond the first agent.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsApplication and API credentials
List static API keys, database passwords, signing keys, webhook secrets, certificates, personal access tokens, deploy keys, and SSH keys. Note where each secret is stored and every legitimate consumer that must receive its replacement.
Delegated and tool access
Inventory OAuth consent grants, OIDC trust relationships, SSO sessions, tool or MCP server credentials, repository integrations, CI/CD connections, and downstream data-service permissions. Record scopes and audiences, not just provider names. Two grants to the same service may have very different read, write, or administrative authority.
Stop the agent without destroying evidence
Stopping execution prevents that process from issuing new tool calls, but it does not revoke credentials copied elsewhere. Use a staged shutdown so investigators retain useful evidence.
Rank #2
- 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.
- Pause queues, schedules, event triggers, and webhooks that can start another run.
- Quarantine the worker or runtime from sensitive networks. Block outbound access to tool servers, repositories, cloud APIs, databases, and secret-management endpoints while preserving a controlled forensic path.
- Prevent automatic restarts and disable deployment automation that could recreate the workload with the same identity.
- Capture volatile details such as active sessions, process metadata, environment configuration, mounted secrets, network connections, and recent tool calls where your incident procedures permit.
- Keep a read-only copy for analysis and label any copy that may contain credentials as sensitive incident evidence.
If the runtime is still making destructive calls, prioritize issuer-side denial and network containment over a perfect snapshot. Document what was lost and the exact cutoff time.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Revoke the workload or service identity at the issuer
The issuer controls whether an identity is accepted. For a Google service account, Google guidance identifies disabling or deleting the account, or applying an IAM deny policy, as the controls for blocking the principal while investigating. Use the equivalent principal-level control at another cloud or SaaS provider.
Why deleting a key is not enough
Deleting a service-account key prevents future authentication with that key, but short-lived credentials already issued from the service identity may remain valid. If immediate cutoff is required, disable, deny, or delete the represented service identity. Also remove permissions that let an unauthorized principal impersonate the account or mint tokens, including the Service Account Token Creator role where it applies.
Use the narrowest effective issuer control
A deny policy can be preferable when you need a reversible emergency block and must preserve the identity for forensics. Disabling is a strong cutoff when the account should not authenticate during the investigation. Deletion is harder to undo and may remove context needed for evidence, so use it when policy or the provider’s incident procedure requires permanent retirement.
After applying the control, test from a controlled environment that new authentication fails and monitor for attempts to recreate credentials through token-minting, impersonation, or deployment roles.
Recommended Free Tools
Revoke OAuth, SSO, and tool-server access
Delegated authorization is separate from the agent’s runtime identity. Revoke every grant at the provider that issued it, and repeat the action for every connected service.
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
OAuth and OIDC grants
Remove the application’s consent grant and any refresh-token sessions. If the provider distinguishes user grants from organization-wide administrator approval, revoke both where appropriate. Invalidate active SSO sessions and trust relationships that can silently issue another token. Narrow scopes on replacement grants to the exact resources and operations required.
Source control and CI/CD
Revoke personal access tokens, deploy keys, SSH keys, repository webhooks, app installations, and CI/CD variables. GitHub Enterprise supports bulk and targeted deletion or revocation for these credential classes; use the option that covers every affected repository and account, then inspect workflow history and recent changes. Search for newly added runners, workflow files, deploy keys, and secret-exfiltration jobs before restoring automation.
Tool and MCP connections
Disable the agent’s credentials at each tool server, not merely in the orchestrator. Remove server-side sessions, rotate shared tool secrets, and review tool authorization rules for read, write, administrative, and credential-minting operations. A tool server may retain its own refresh token even after the agent process stops.
Rotate persistent secrets safely
Rotation is a controlled replacement, not a single delete action. Follow this order:
- Block the old path. Apply issuer denial, network controls, and delegated-grant revocation first so an attacker cannot use the old secret while the replacement is being deployed.
- Create a new secret with less authority. Use a distinct identity, narrow scope, short expiration, and an audience restriction. Do not copy the old roles by default.
- Update every legitimate consumer. Replace the secret in the agent, CI/CD jobs, tool servers, data connectors, and deployment manifests. Check replicas, failover regions, and scheduled jobs.
- Test the replacement. Verify required reads, writes, and tool calls from a controlled test run. Confirm that prohibited operations fail.
- Disable and delete the old credential. Remove old key material from secret stores, build logs, images, caches, and configuration backups according to retention policy.
- Watch for residual use. Alert on authentication attempts with the retired key, calls from unexpected locations, and attempts to mint or impersonate credentials.
Rotate signing keys and certificates when they could have been exposed. Update verifiers and trust stores in a coordinated window so legitimate services do not fail while old signatures are being retired.
Investigate what the agent actually did
Do not rebuild immediately. First determine whether the agent read, changed, or exfiltrated information.
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.
Authentication and identity evidence
Review service-account usage, key IDs, token issuance, impersonation events, OAuth consent changes, SSO sessions, and failed authentication after the cutoff. Correlate timestamps with runtime logs and deployment events.
Tool, network, and data evidence
Examine every tool call, command argument, repository operation, database query, object-storage access, and outbound connection. Look for unusual destinations, bulk reads, privilege changes, downloads, and attempts to access secret stores.
Integrity and persistence checks
Compare repositories, infrastructure definitions, workflow files, container images, schedules, webhooks, and access-control lists with known-good versions. Search for newly created users, keys, service accounts, runners, and automation that could restore the attacker’s access.
Preserve evidence with timestamps and chain-of-custody controls appropriate to your organization. The Australian Cyber and Infrastructure Security Centre and Google Cloud both emphasize retaining logs and examining affected resources before restoration.
What “revoked” actually means
| Action | What it stops | What it does not guarantee |
|---|---|---|
| Delete a persistent API or service-account key | Future authentication attempts using that key | Already-issued short-lived tokens may continue to work. |
| Disable, deny, or delete the service identity | The principal’s authentication and, where supported, active-token use | Other credentials or delegated grants held by the agent. |
| Revoke an OAuth or SSO grant | Delegated access through that provider | Access through another connected service or a copied credential. |
| Stop the runtime | New calls from that process | Tokens, keys, sessions, or copies used elsewhere. |
| Invalidate an authenticator | The compromised authenticator | Separate authenticators or identities that were also exposed. |
NIST SP 800-63B states: “CSPs SHALL promptly invalidate authenticators when … the authenticator is compromised.” Apply that principle at the issuer that can actually invalidate the credential, rather than assuming a local process stop is equivalent to revocation.
Rebuild the agent with a smaller blast radius
Restore service only after the investigation establishes what must be trusted. Create a new workload identity instead of reusing the compromised one.
Best Value
- FIDO2 & Passkey Ready: Business-ready and FIDO2 L1 certified. This key is supported by major management suites and is ideal for both individual and enterprise deployment. Works seamlessly with Gmail, Facebook, GitHub, Dropbox, Coinbase, and more.
- Universal Connectivity (USB-A ): Features a built-in USB-A connector—simply unfold the key and plug it into your compatible PC or laptop for seamless authentication on the go.
- Dedicated Manager App: Use the Thetis Manager App for the initial hardware PIN setup. Setting the PIN on the device first ensures a smooth registration process. Once the PIN is configured, you can begin registering the key across your favorite FIDO2-compatible online services.
- Ultra-Durable & Portable: Featuring a rotating metal cover, this key is water, crush, and tamper-resistant. It fits easily on a keychain and requires no batteries or network connectivity.
- Check FIDO2 compatibility before purchase - Known limitations: ID Austria is not supported (requires FIDO2 Level 2). Windows Hello login only works with Windows Enterprise editions that support Entra ID, and NFC is NOT supported.
Use one identity per agent
Give each agent and environment a distinct workload identity. Never share a human administrator’s broad credential across agents, and do not let unrelated agents share a service account. Separate development, testing, and production identities.
Prefer short-lived, scoped credentials
NIST describes long-lived API keys and bearer tokens as broad bearer credentials. Prefer dynamically issued credentials that are task-scoped, audience-restricted, and short-lived. Set the audience to the intended service, limit resources and operations, and require renewal rather than indefinite reuse.
Use standard, verifiable protocols
NIST SP 800-228 recommends cryptographic verification, regular credential cycling, signing-key rotation, and standard mechanisms such as OAuth2/OIDC, SPIFFE SVIDs, and JWTs. Validate issuer, signature, audience, subject, scope, expiration, and nonce or replay protections where applicable.
Separate dangerous capabilities
Keep data reading, data writing, tool invocation, deployment, and credential minting in separate permission sets. An agent that only summarizes documents should not be able to deploy code or create tokens. Require human approval or an independent policy check for irreversible operations.
Add a tested kill switch
The kill switch should disable the workload identity and block tool egress without depending on the compromised agent. Exercise it in a controlled drill, measure the time to cutoff, and document who can activate it.
Choose controls by the protection they provide
| Control | Cutoff speed | Token coverage | Typical blast radius | Best use |
|---|---|---|---|---|
| Pause or stop the runtime | Fast for that process | None outside the process | Single runtime | Preventing immediate new tool calls while preserving evidence |
| Network egress block | Fast when centrally enforced | Blocks reachable services, not copied credentials elsewhere | Worker, subnet, or environment | Containing tool and data access |
| Delete a persistent key | Fast for future use | Does not necessarily cover issued short-lived tokens | One key | Retiring exposed static material after a stronger cutoff |
| Disable or deny the workload identity | Fast at the issuer | Designed to stop the principal and active-token use where supported | Every workload using that identity | Immediate identity containment |
| Revoke OAuth or SSO grants | Provider-dependent | That delegated provider and grant | All clients using the grant | Removing SaaS and user-delegated access |
| Rotate and replace secrets | Requires deployment time | Old secret after disablement | Every legitimate consumer of the secret | Recovery and long-term remediation |
Evaluate each control by cutoff speed, coverage of already-issued tokens, scope, blast radius, auditability, recovery time, and support for short-lived audience-restricted credentials. In a live incident, combine controls rather than waiting for one mechanism to solve every path.
Protect the human administrator performing recovery
The operator account used to revoke credentials can become the next target. Require phishing-resistant authentication and review its sessions, recovery methods, roles, and recent activity before using it for containment.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Repair Windows errors before they cause bigger problemsFix Now →Scan for outdated or missing drivers - takes under a minuteDriver Scan →A FIDO2 security key is a relevant hardware authenticator for that administrator account. It protects the person carrying out revocation and recovery; it does not revoke agent tokens or disable a service identity by itself. Perform the actual revocation in the cloud, identity, SaaS, repository, and tool issuers.
Operational checklist for closure
- Every affected runtime, scheduler, webhook, service account, API key, OAuth grant, tool credential, repository integration, and downstream data connection is recorded.
- Logs and forensic copies are preserved with timestamps.
- The runtime is paused or quarantined, and sensitive egress is blocked.
- The underlying workload identity is disabled, denied, or retired.
- Delegated grants, SSO sessions, personal tokens, deploy keys, SSH keys, and tool-server credentials are revoked.
- Impersonation and token-minting roles are removed or constrained.
- Replacement credentials are short-lived, scoped, audience-restricted, and issued to a unique workload identity.
- Repositories, workflows, infrastructure, data access, and identity records show no unauthorized changes.
- Post-cutoff authentication and credential-recreation attempts are monitored.
- The kill switch has been tested, owners are assigned, and the restoration decision is documented.
Containment is complete only when the agent’s identity, delegated authorizations, copied secrets, and reachable systems have been considered together. Restoration is a separate decision, made after evidence review and a least-privilege rebuild.
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.
