Free tools Windows power users keep installed
One-click scans. No signup required.
Put safety controls around the agent, not just inside its prompt. Let narrowly scoped, low-risk actions run automatically, and require an external authorization check or human approval when a tool call could cause significant harm, change important data, or communicate outside the system. The agent can then continue routine work without being trusted to decide the limits of its own authority.
Why an agent can take an unsafe action
An agent may read an email, file, or webpage that contains instructions written by someone other than the user. If the agent treats that content as instructions rather than untrusted task data, it may be manipulated into taking actions the user never requested. NIST describes this as agent hijacking through indirect prompt injection and highlights the difficulty of reliably distinguishing trusted instructions from untrusted data.
As an Amazon Associate I earn from qualifying purchases.
This is why a system prompt, refusal behavior, or keyword filter cannot guarantee safety. Those measures can help, but they do not create a dependable boundary at the moment an action changes data or reaches another person or system. That boundary belongs in the software that executes the tool call, or in the downstream system that accepts it.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesBuild the controls around the action boundary
1. Map what the agent can do
List the tools the agent can call, the data each tool can reach, and the external destinations it can contact. Classify operations by their likely impact, not merely by the name of the tool. OWASP gives an illustrative—not universal—risk scheme: document search and file reading are low risk; file writing is medium; email sending and code execution are high; database deletion and money transfers are critical.
#1 Best Overall
- Ultra-Compact FIDO2 Security Key - Plug-and-stay or carry on a keychain. This USB-A hardware security key offers portable, always-on protection for desktop and mobile use. (Item Size: 0.75 X 0.74 IN x 0.25 IN)
- USB-A Hardware Key for All Devices - Works with USB-A ports on PC, Mac, Android, and other laptop/notebook device. Enables secure, cross-platform login with FIDO2.0 passkey support.
- FIDO Certified Security Key - Meets FIDO and FIDO2 standards. Works with Google, Microsoft, GitHub, Dropbox, and more. Please check service compatibility before purchase.
- Passwordless Login with Passkey - Supports passkey login via WebAuthn and CTAP2. Enjoy password-free sign-ins where supported. Not all websites or services currently support passkeys.
- Advanced Multi-Factor Authentication - Offers 200 FIDO2 passkey slots and 50 OATH-TOTP slots. Strong, flexible 2FA/MFA support across various apps and authentication platforms.
Use the classification to decide which operations may run without review and which must stop for authorization. A tool’s risk can change with its target or arguments: reading a public document is different from reading a sensitive record, and writing a draft is different from overwriting a production file.
2. Give each task only the capabilities it needs
Remove unused tools and split broad tools into narrow operations. If a task is to summarize email, give the agent a read-only operation rather than a general email tool that can also send or delete messages. Where practical, keep read-only and write-capable identities separate.
Rank #2
- Packing List: This doorbell removal tool set is made of high-quality metal and comes in four types and comes with two doorbell removal pins and a key ring. These kits can be hung on a key ring, making them portable and loss-proof.You will get: 8 x Security Pin Key Release Removal Tool,1 x key ring.
- Anti-slip Handle Design: It has a solid and anti-slip handle, which is easy to grasp and saves effort when using it.
- Wide Application: It could be used for replacing your lost security key to remove your Nest Hello, Arlo and Eufy Video Doorbell from its mount.It can even be used to detach part of the metal watch strap.
- Compatibility: Fits various models of video doorbell. All Arlo Video Doorbell Models, all Eufy Video Doorbell models, and all Nest video doorbell models.
- Multi Usages: With this tool, you could replicate the action of the manufacturer security pin but inserting it on either the top or bottom, dependent on model and pulling gently on the doorbell to release it.
Apply least privilege to data and destinations as well as operations. An agent that needs one project folder should not receive access to an entire drive; a task that needs to post to one approved destination should not be able to contact any destination. OWASP’s DevSecOps guidance calls this “least agency”: give the agent only the autonomy, tools, and access its task requires, for only as long as it needs them.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
3. Authorize every consequential call outside the model
At execution time, check the caller, resource, operation, and arguments against policy. Enforce that check in the tool, execution service, or downstream system—not by asking the model whether it is allowed. A natural-language instruction or a model-generated “approved” flag is not authorization.
Rank #3
- HARDWARE 2FA AND MFA: FIDO Alliance Certified FIDO2 v2.1 with CTAP2 plus legacy U2F and CTAP1 for strong two-factor login and passwordless sign-in on services that support security keys
- BUILDING ACCESS ON ONE CARD: MIFARE DESFire EV2 4K applet with AES encryption adds office door and physical access control alongside digital authentication
- CERTIFIED SECURE ELEMENT: An NXP Common Criteria EAL6+ certified secure controller and Java Card platform protects your keys on a tamper-resistant chip
- DUAL INTERFACE SMART CARD: Contactless NFC ISO 14443 plus ISO 7816 contact reader support in an ISO 7810 ID-1 format that is passive and needs no battery
- SWISS ENGINEERED DESIGN: Built by Cryptnox as a single card for authentication and access control and backed by a 2 year warranty
Validate arguments at the same boundary. Reject requests that name an out-of-scope target, request an unpermitted operation, or contain parameters the tool does not accept. For critical authorization failures, fail closed: do not perform the action while the policy check is unavailable or inconclusive.
4. Ask for review when the impact warrants it
For high-impact or destructive operations, present a reviewer with the proposed action and its actual arguments. Bind the approval to the exact actor, tool, target resource, parameters, time, and expiry. If any material parameter changes after approval, require a new decision rather than reusing the old one. Add replay protection and idempotency where appropriate so that retries cannot silently repeat a consequential action.
Rank #4
- A FIDO security key with PUF technology provides a unique, hardware-rooted trust anchor that resists tampering and cyber attacks, offering stronger security than conventional designs.
- FIDO2 Certified Protection – Enjoy phishing-resistant security with FIDO2 certification, ensuring top-tier account safety across Windows, macOS, Linux, iOS iOS, Android and more.
- Easy to use & Portable – Designed with a compact USB-C interface, Clife key fits easily on your keychain for secure access anywhere. Simply plug in and authenticate with ease.
- Universal Compatibility – Works seamlessly with hundreds of FIDO2/U2F compliant services, including popular cloud, email, and social platforms.
- Backup recommended – To ensure continuous access, register a backup Clife security key as a spare in case your primary key is lost.
Keep the workflow resumable. Store the pending action and its context, pause only that operation for review, and continue the task after approval by revalidating the authorization immediately before execution. If approval is denied or expires, return a clear status to the agent so it can stop, ask the user for another path, or complete other safe parts of the task. Do not silently substitute a different target or operation.
Keep prompt-injection defenses as supporting layers
Separate trusted instructions from external content in the agent’s context, label untrusted material where possible, and use structured prompts or filters as additional safeguards. OWASP cautions that such measures are not a complete defense against prompt injection. Malicious text can arrive through ordinary task data, so the decisive protection is still to constrain and authorize the action where side effects occur.
Best Value
- Protect accounts with USB-A & NFC 2FA security key. Hardware-based authentication blocks phishing, credential theft & unauthorized access across cloud, enterprise & personal platforms.
- FIDO2 Level 2 certified Security Key. TAA compliant and supports Apple ID, Microsoft Azure/Entra ID, AWS, Google, Facebook, Salesforce, DUO & more. Works with Chrome, Safari & Edge across major OS.
- Plug & play USB-A Security Key with NFC tap login. No software, drivers or batteries required. Works with Windows PC, MacBook, iPhone, Android & Chromebook for fast, secure authentication.
- Built with FIPS 140-2 Level 3 secure element for advanced encryption. Trusted by IT teams, healthcare, education & government for secure authentication and identity protection.
- IP68 waterproof, dustproof & crush-resistant design. Supports FIDO2, U2F, OTP, PIV, Mini Driver & smart card login. Durable USB security key for long-term enterprise and daily use.
Reduce the consequences of a missed check with task-scoped identities, short-lived credentials, sandboxing, and monitoring. Restrict the sandbox’s shell, filesystem, network, and tool integrations to what the task requires; isolation is only useful to the extent that it covers the paths the agent can actually use. Keep production data and credentials out of an agent’s reach where feasible, and log policy decisions and action outcomes for review.
Test safety and workflow continuity together
A safeguard that blocks an attack but also refuses every legitimate task is not a good workflow control. Evaluate both whether unsafe actions are stopped and whether the intended task can still be completed. NIST recommends adaptive, task-specific evaluation; its guidance also notes that repeated attempts may provide a more realistic evaluation than a single try. This is qualitative guidance, not a quantified estimate of real-world risk or safeguard effectiveness.
- Set a benign control task. Choose a normal request that should complete within the agent’s permitted scope.
- Test direct and indirect attacks. Include an attempt in the user’s request and place another in the external content channel being tested, such as a dummy email, file, or webpage.
- Use safe test infrastructure. Substitute dummy data and instrumented tools so attempts to send, delete, execute, or modify something cannot affect real users or systems.
- Record two outcomes separately. Track whether policy blocked the prohibited action and whether the legitimate task completed. Also inspect false refusals, unexpected tool calls, and whether a reviewer could understand the proposed action.
- Repeat after changes. Re-run relevant cases when tools, permissions, prompts, policies, or approval paths change, using the deployed configuration rather than assuming a policy works because it exists on paper.
Choose controls by the boundary they protect
When assessing an implementation, compare where its authority comes from and how it behaves when a decision is needed—not just whether it advertises an approval step or sandbox.
Recommended Free Tools
- Scope: Are tools, data, identities, and destinations restricted to the task?
- Enforcement: Does an external component authorize every consequential action and validate its arguments?
- Approval: Is review tied to the exact proposed action, and can the workflow safely resume after the decision?
- Containment: Does isolation cover the shell, filesystem, network, and tool integrations the agent can reach?
- Evaluation: Are attack blocking and legitimate task completion both measured, including false refusals?
OWASP and NIST provide general security guidance, not proof that any particular framework, model, policy engine, or product will prevent unsafe actions. Permissions, side effects, approval paths, and sandbox coverage depend on the implementation, so assess the actual deployed configuration with realistic but harmless inputs and instrumented 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.
