What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Give the agent a dedicated, accountable identity; limit it to the data and actions needed for a specific task; and enforce authorization at the connector and the downstream service—not just in the prompt. Require fresh human approval for consequential changes, and keep action-level logs and a tested way to revoke access. These controls reduce the impact of mistakes or misuse; they cannot eliminate prompt injection or other agent risks.
Start by defining the agent’s job and security boundary
Before connecting an agent, document what it is for, who owns it, where it runs, and which users, data stores, connectors, tools, and downstream services are involved. Specify the task it is allowed to perform and what it must never do. Note whether it acts with a user’s delegated authority or under its own autonomous identity, and include guest or cross-tenant paths where relevant.
Think of the agent as a principal with an owner, purpose, identity, permissions, tool set, and lifecycle. The security boundary must be enforced by ordinary identity and access controls at the tool, connector, and downstream service. A prompt can guide behavior, but it is not an authorization mechanism.
Give the agent a distinct identity and task-scoped access
Assign each agent a unique, lifecycle-managed identity with a named owner. Avoid a broad shared service account or shared secret used by unrelated agents: it obscures accountability and can give a compromised or misdirected agent a wider path into company systems. Keep credentials out of prompts and other model-visible text; have the trusted runtime or connector handle authentication.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstall#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.
Grant only the permissions necessary for the defined task. Scope access by resource, tenant or workspace, data sensitivity, and operation. Review the agent’s effective access across its roles, integrations, and downstream systems together; individually narrow grants can combine into broad access.
Where practical, separate retrieval from editing, sending, exporting, deleting, deploying, and changing permissions. Use short-lived credentials or just-in-time elevation for temporary privilege, rather than leaving elevated access in place. Microsoft Learn’s current guidance on least privilege for AI agents recommends owned identities, task-scoped roles, permission review, and revocation testing.
Expose only reviewed tools, and authorize every call
Allowlist the tools and actions the agent can invoke. A connector should not expose a capability merely because the model might find it useful. Give read-only retrieval a separate path from operations that change data or affect other people, and put high-impact actions—such as deletion, external sharing, purchases, deployments, or access changes—behind a fresh human approval or time-bound elevation.
Rank #2
- 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
For each request, the connector or tool server should authenticate the caller and check whether that identity may perform that specific action on that specific resource. The downstream service must also enforce its own authorization. Do not assume that signing in to an agent interface protects an API or data store it can reach.
Recommended Free Tools
If the connector uses MCP
Treat a Model Context Protocol (MCP) server like an API. Microsoft Learn’s guidance, “Secure a Model Context Protocol (MCP) server with Microsoft Entra ID,” says to require an OAuth 2.0 access token on every request and validate it before running a tool. Bind the request to the initiating principal and the requested scope, then verify that the MCP server and downstream system both enforce that identity’s actual permissions. Confirm the supported OAuth flow and permission behavior in the documentation for the exact product and configuration you deploy.
Assume connected content can be hostile
Documents, emails, web pages, user inputs, tool descriptions, and tool outputs can contain malicious instructions. Direct or indirect prompt injection may try to steer an agent into exposing information or invoking a tool improperly. OWASP’s “AI Agent Security Cheat Sheet” discusses prompt injection and tool abuse and recommends least privilege.
Rank #3
- 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
Use input and output checks and prompt-injection defenses as additional safeguards, but keep access decisions in deterministic authorization logic outside the model. Review connector configuration and tool descriptions, since they shape what the agent can call. If untrusted content persuades the model to request an action, the tool boundary should still reject an unauthorized request.
Choose an identity and capability model deliberately
| Design choice | What it means | Security question to answer |
|---|---|---|
| Dedicated agent identity | The agent has its own named, lifecycle-managed principal. | Can you identify its owner, review its effective access, and revoke it without disrupting unrelated agents? |
| Shared service credentials | Multiple agents or workflows use the same account or secret. | Can you reliably attribute an action and limit the effect of a credential exposure? If not, avoid this design. |
| User-delegated execution | The agent acts with authority associated with an initiating user. | Is the user’s identity and scope carried through every tool call and checked downstream? |
| Autonomous execution | The agent acts under its own identity rather than a user’s delegated authority. | Are its independent permissions bounded to one task, and are consequential actions approved? |
| Read-only retrieval | The agent can retrieve data but cannot change it. | Are the sources, tenant boundaries, and sensitivity levels no broader than the task requires? |
| Write or administrative actions | The agent can change, send, delete, deploy, export, or alter access. | Are actions individually authorized, logged, and subject to fresh approval or temporary elevation when high impact? |
For vendor or deployment comparisons, also check scope precision, per-call authorization, tool allowlists, approval controls, audit quality, practical revocation, data residency, tenancy, identity-provider compatibility, and security-monitoring integration. Validate the exact version, configuration, and contractual data handling; general guidance does not establish that a particular product or setup has these capabilities.
Log actions and rehearse containment
Keep enough evidence to reconstruct what happened across the orchestrator, connector, and service. At a minimum, record the agent and caller identity where relevant, role or scope, tool and action, target resource, authorization outcome, and a correlation ID that links the events. Protect these records under the organization’s logging and retention controls.
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.
Test containment before relying on the integration. Rehearse disabling the agent, rotating its credentials, invalidating or expiring tokens, and removing downstream grants—including stale permissions that remain after a connector is changed. Re-review effective access when the data scope, tools, owner, or deployment changes.
Roll out in stages
- Inventory: Record the workflow, owner, environment, data sources, connectors, tools, downstream services, identity model, and prohibited actions.
- Design identity and scope: Create a dedicated principal, map its task-specific roles, and check aggregate access across integrations.
- Constrain capabilities: Publish only reviewed tools; separate retrieval from changes; require fresh approval or temporary elevation for high-impact operations.
- Enforce each hop: Validate identity and scope at the tool boundary and again in downstream services. For MCP, validate an OAuth 2.0 access token before tool execution.
- Test bounded behavior: Start with low-risk, read-only work in a limited environment. Test representative benign and adversarial inputs, including indirect prompt injection and attempts to trigger unauthorized actions.
- Review before expanding: Inspect logs and effective permissions before adding data sources or write actions. Rehearse revocation and credential rotation, then repeat the review after material changes.
Protect the administrator account as well
Secure the human accounts that configure the agent and its identity provider. CISA’s guidance describes hardware-based FIDO keys as a phishing-resistant sign-in option where supported. A key can strengthen a human administrator’s sign-in; it does not constrain the agent’s permissions or authorize its tool calls.
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.
Free tools Windows power users keep installed
One-click scans. No signup required.
