iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Require approval at the tool boundary, immediately before an AI agent causes a side effect. The model may propose an email, record edit, shell command, purchase, publication, or deletion, but your application—not the model—decides whether that call is allowed to run. If review is required, pause the run, show a person the exact stored tool name and arguments, authenticate the reviewer, consume the pending request once, and then resume the original run with an approve or reject decision. This design works across OpenAI Agents SDK, LangChain/LangGraph, Microsoft Agent Framework, and n8n, although each framework persists and resumes paused work differently.
What does human approval for an AI agent action mean?
Human approval is a policy checkpoint between an agent’s proposed tool call and the tool’s execution. The model emits a structured request; application code validates it, decides whether review is necessary, and invokes the operation only after a valid decision. As Anthropic describes the contract, the model does not run anything itself: it emits a request, your code or the service performs the operation, and the result is returned to the conversation.
This boundary matters because a harmless-looking instruction can produce an irreversible effect. Sending a message, changing a customer record, running code in production, disclosing personal data, publishing content, or deleting files all deserve a decision that is independent of the model’s text generation.
Which agent actions should require a person?
Start with an inventory of every tool and classify the consequence of a mistake. A practical policy separates read-only work from operations that alter external state.
#1 Best Overall
- Powerful Turbo Fan:CAGIWIRU Electric Air Duster have reached speed up to 130000RPM,Can efficient remove dust and debris.With LED light and Three Adjustable Speed to meet with different cleaning tasks.
- With Attached USB C Fasting Charging cable.Charing 3 hours,Enjoy up to 240 minutes of use on the lowest setting, For third speed can use for 25mins.with four charging options to meet with your needs.
- Wide Application: attached 5 different nozzles,making it suitable for a variety of scenes, such as car,pet, pc, keyboards, and other electronic devices. It also serves for office and home clean duster.
- Handy Design & Portable:Cordless air duster electric nozzle is Light weight and small size, design and comfortable to hold and space saving, convenient to carry around. You can put it in your room or inside the car without taking up space.
- Economical &Reusable: Built durable materials with a long-lasting battery,our CAGIWIRU air duster it's a sustainable alternative to disposable air cans.
- Low risk: read-only searches, calculations, draft generation, or retrieval from non-sensitive data. These can usually run without an interruption when access controls are already enforced.
- Medium risk: edits to ordinary records, outbound messages to known recipients, or changes that can be reversed. Require approval when the data, recipient, or business rule makes an error costly.
- High risk: deletion, financial transactions, purchases, production shell commands, publication, permission changes, access to sensitive data, or disclosure outside the organization. Make review mandatory and consider a second approver.
Document the risk level beside the tool definition, not only in a prompt. Prompts can be ignored or misunderstood; a tool-bound rule is evaluated by the application that has authority to execute the side effect.
How does an approval-gated agent run work?
The flow has six distinct states: proposal, validation, interruption, decision, atomic consumption, and execution. Keeping those states separate prevents a browser request or a model message from becoming an implicit authorization.
- The model emits a structured tool call containing the tool identity and arguments.
- The host validates the schema and evaluates the tool’s approval policy. Malformed arguments must not be executed.
- If review is required, the host stores the pending call and pauses the run before invoking the tool.
- The review client displays a redacted record and submits only an opaque decision identifier plus an approve, edit, or reject choice where the framework supports it.
- The server authenticates the reviewer, authorizes access to that run, verifies that the request is still pending, and consumes it atomically.
- Only the server-owned snapshot is resumed. An approval executes the stored call; a rejection becomes a controlled result to the agent or ends the workflow.
OpenAI’s guidance summarizes the intended behavior: the model can decide an action is needed, but the run pauses until a person approves or rejects it. The same pattern applies whether the run is streamed, nested inside another agent, or connected to an external tool protocol.
Quick wins for a faster PC:
Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →How do I add approval step by step?
1. Inventory side effects and assign policy
List every function, API wrapper, MCP action, workflow node, and shell or database operation the agent can reach. Record the data it can touch, whether the operation is reversible, who can be affected, and the maximum plausible loss from an error. Mark each tool as always requiring approval, conditionally requiring approval, or never requiring approval. Revisit the list whenever a tool, destination, or permission changes.
2. Declare the rule on the tool
Use the framework’s native gate where possible. In OpenAI Agents SDK, needsApproval can be set to true or to a function that evaluates the call; hosted MCP tools use require_approval. Microsoft Agent Framework exposes approval_mode="always_require" for function tools. LangChain/LangGraph applies review policy through HITL middleware, while n8n places an approval gate in the workflow. A rule attached directly to the executable tool is harder to bypass than an instruction placed only in the agent prompt.
3. Stop before invocation
When a gated call is emitted, do not invoke the function and do not silently substitute a safer-looking call. Store the tool name, parsed arguments, run identity, policy version, creation time, and expiration time, then return an interruption. OpenAI exposes pending ToolApprovalItem interruptions; LangChain’s middleware creates an interrupt containing action requests and review configuration.
Validation belongs before the pause as well as before execution. If a JSON argument is malformed, incomplete, or outside the tool schema, route it to correction or rejection. OpenAI’s SDK requests approval for malformed arguments rather than invoking the approval callback or executing the tool, which prevents a broken payload from slipping through an approval path.
Free tools Windows power users keep installed
One-click scans. No signup required.
4. Build a review record, not a raw model transcript
The reviewer needs enough context to make a decision without being asked to read the entire conversation. Display:
Rank #2
- Screen Cleaner is Your Screen’s New Bestfriend! - EVEO is proud to present a screen cleaner that’s perfectly suited for your TV. Meet your new screen cleaner, computer screen cleaner, laptop cleaner, iPad cleaner, Macbook,tv cleaner or any other electronic device. Our screen cleaner spray and wipe along with the included plush microfiber cloth, easily removes dust, fingerprints & other smudges on the screen’s surface. Clean effectively for a streak-free screen
- Gentle on your Screens - Our screen cleaner spray and wipes is compatible to be used even for the most sensitive LCD, LED, CRT, and OLED screens in proper use. Be at ease knowing that this product is compatible with all the major brands: LG, SONY, Samsung, Sharp, iPad, iPhone, Android screens, Kindle, PC monitors - we’ve got you covered! Can be used as LCD screen cleaner, tv cleaner, smart tv screen cleaner, and more
- Premium Super Soft Plush Microfiber Cloth - Included in this efficient screen cleaning kit. This screen cleaner spray and wipe has our signature plush microfiber cloth specially designed to comply with screens and monitors and leave them streak-free. Has been shown to effectively absorbs stubborn stains, zero streaks, and no fiber shedding. Always use this together with our screen cleaner for the best results
- Easy-to-Use Screen Cleaner Kit + microfiber cleaning cloth - The EVEO Screen Cleaning Kit will easily be your new favorite tool when it comes to taking care of the things you love most, quick and easy solution for a streak-free screen. Simply screw on the sprayer gun on the screen cleaner bottle. Spray the cleaning liquid onto the microfiber cloth, then wipe the screen until desired cleanliness & shine is achieved.
- Quality Screen Care Kit Worth Purchasing - Enjoy a convenient Screen Cleaner anytime, anywhere you need it. Your satisfaction is our top priority. We stand behind the quality of our products and that’s why, you shouldn’t be worrying at all. We manufacture the Best Screen Cleaner. and can be used for laptop cleaning kit, laptop screen cleaner, tv screen cleaner for smart tv. MacBook screen cleaner, monitor cleaner or screen cleaner spray for electronic devices with microfiber cleaning cloth
- the plain-language action and the exact tool identity;
- all parsed arguments, including target account, record, path, recipient, amount, or command;
- the expected side effect and whether it is reversible;
- the user, agent, run, environment, and originating request;
- the policy that triggered review and its version;
- the time the request was created and when it expires;
- the available choices and the consequences of each.
Treat tool names, arguments, and explanatory text as untrusted model-controlled content. Escape them before inserting them into HTML, and redact credentials, access tokens, unnecessary personal data, and secrets. Keep the complete state on the server; expose only the minimum fields needed for the decision.
5. Authenticate and authorize the reviewer
The browser should submit an opaque decision ID and a choice, not a replacement tool call or a new set of arguments. On the server, authenticate the session, check that the person is allowed to approve that class of action, load the stored run, and verify that the decision ID belongs to a still-pending request owned by that run. A run ID, approval ID, or URL token is an identifier—not proof of authorization.
Use least privilege for reviewers as well as agents. Someone allowed to approve routine CRM edits should not automatically be able to approve production shell commands or disclosure of regulated data. Where separation of duties is required, enforce it server-side rather than relying on a second button in the interface.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errors6. Consume the decision atomically
Approve, reject, or edit must be a conditional state transition such as “pending” to “approved” or “rejected.” Perform it in a transaction or with an equivalent compare-and-set operation. If two browser tabs submit the same decision, only one may win; the other must receive an already-consumed or expired response and must not resume the run again.
7. Resume the original run
Apply the decision to the server-owned paused state and resume the original top-level agent. Preserve the same session when using streaming or conversational memory so the agent receives a coherent tool result or rejection. OpenAI Agents SDK serializes paused execution in RunState; LangGraph persists graph state; Microsoft Agent Framework leaves continuation ownership to the caller; n8n pauses the workflow execution until its approval gate is resolved.
8. Define rejection and expiry
A rejection can end the workflow, return a structured denial to the model, or ask the model to propose a safer alternative. OpenAI supports custom rejection messages, and LangChain can synthesize a tool message for a rejected call. Choose one behavior per tool so the agent cannot interpret a denial as successful execution.
Set an explicit expiration. When the deadline passes, mark the request expired and require a fresh proposal. Never treat silence, a closed browser, or a worker restart as approval. A new proposal should be revalidated because the target, permissions, or policy may have changed while the old request was waiting.
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 →9. Write an audit trail
Record the proposed call, normalized arguments, reviewer identity, decision, timestamp, policy version, expiration, and resulting tool output or error. Keep full run state and diagnostic errors on the server while exposing a restricted audit view to ordinary users. Audit records should make it possible to answer who approved what, under which policy, and whether the operation actually succeeded.
Rank #3
- 【Large size】：Measuring 800 x 300 mm/ 31. 5×12 inches, which is wider and taller than others. Large mouse pad can perfectly fit your keyboard and mouse of any size, and there is also ample space for peripherals such as phones, tablets, etc.
- 【LED backlight mouse pad】：Blade Hawks RGB mouse pad provides 7 static modes and 7 dynamic modes will give you the options you are looking for. Power-off memory function remembers the last lighting mode you selected. The super glow fiber Chroma will give you a colorful gaming setup that you want.
- 【Anti-slip rubber base】：Made from Natural rubber, this gaming mousepad will firmly grip your desktop, allowing you to enjoy the ultimate precision in the games you are most passionate about. It also comes with a texture that enhances this ability even further.
- 【Smooth and waterproof surface】: Micro-textured cloth surface, for extremely precise mouse control and a smooth movement. When liquid splashes on the gaming mouse pad, it forms water droplets and slides down. Will not delay your work or gameplay.
- 【Plug and play】: This mouse and keyboard pad is powered by USB, plug and play, no driver required. One button control light modes. Press one time to change the lighting mode, press twice to change the brightness, press and hold about 3 seconds turn ON/OFF.
What should an AI action approval screen display?
An effective screen lets a reviewer decide from a stable, human-readable record while preserving the exact call that will execute. Put the proposed effect first, then the evidence and controls.
| Screen area | What to show | Why it matters |
|---|---|---|
| Action summary | Verb, tool name, target, environment, and a one-sentence consequence | Prevents a technical function name from hiding the real effect |
| Arguments | Parsed key-value fields or a syntax-highlighted read-only view | Lets the reviewer verify recipients, records, paths, amounts, and filters |
| Origin | Requesting user, agent, run identifier, and triggering instruction | Provides accountability and detects an unexpected workflow |
| Policy context | Risk level, rule name, policy version, and required approver role | Explains why review is required |
| Timing | Created time, expiration time, and whether the request is already claimed | Stops stale approvals and clarifies urgency |
| Controls | Approve, reject, and—only where authorized—edit; require confirmation for irreversible actions | Makes the decision explicit and auditable |
Do not allow arbitrary HTML, links, or scripts from model-controlled fields. Display secrets as redacted values and provide a separate, permissioned way to inspect sensitive evidence when it is genuinely necessary.
How do OpenAI Agents SDK approvals work?
OpenAI Agents SDK places approval on the tool with needsApproval; hosted MCP integrations use require_approval. A gated call produces a pending ToolApprovalItem interruption instead of invoking the tool. Your application renders that item, obtains a server-authorized decision, updates the serialized RunState, and resumes the original top-level run.
Use a function for conditional policy—for example, require review only when a message has an external recipient or when a database mutation targets production. Keep the predicate deterministic and based on parsed arguments and trusted context, not on a model-generated explanation. Sticky decisions can apply to later calls when your policy explicitly permits that; do not make a one-time approval broader than the reviewer intended.
OpenAI’s security guidance warns that RunState.fromString() does not authenticate the snapshot or the person submitting it. Verify integrity, ownership, authorization, and replay protection before deserializing or resuming any state supplied through a client-facing flow.
How do I configure approval in LangChain or LangGraph?
Use HITL middleware to interrupt a graph when a configured action request reaches the tool boundary. The interruption carries the requested action and review configuration; persisted graph state allows the application to pause and resume later.
LangChain supports three useful outcomes: approve the exact call, edit permitted fields, or reject it. Treat editing as a new validation step, not as a cosmetic change. After applying an authorized edit, rerun schema, destination, permission, and risk checks before resuming. A rejection can be represented as a tool message so the agent can explain the denial or produce a revised plan without pretending that the original operation ran.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Persist the graph checkpoint in durable storage and bind it to the authenticated user and run. The middleware handles the interruption shape, but your application still owns reviewer authentication, authorization, atomic consumption, expiration, and audit logging.
How does Microsoft Agent Framework require approval?
For function tools, set approval_mode="always_require" when every invocation needs a human decision. The caller collects user input, records the decision, and starts the next run; continuation state is therefore an application responsibility.
Rank #4
- 【Softer and More Comfortable】Vaydeer wrist rest has unique diamond pattern, which is the combination of softness and aesthetics. The materials of wrist rest are improved into higher quality memory foam and covered with silky smooth lycra. The computer wrist rest makes you as comfortable and cushiony as like rest your wrists on clouds.
- 【Ergonomic Wrist Saver】The wrist rests for keyboard and mouse comes with a 17.32×3.15×0.83 inch keyboard wrist pad and a 5.94×3.15×0.83 inch mouse wrist support. Based on ergonomic design, the unique concave shape is the perfect fit for your wrist joints. The wrist rest pad fits most computer keyboards and laptops, improve hand and wrist posture, release your wrist and arm stress.
- 【Non-Slip Rubber Bottom】Featuring an anti-skid silicone base on the bottom, this wrist keyboard support stays firmly in place on your desk, preventing the padding from sliding around, ensuring stable and consistent wrist support during extended computer sessions.
- 【Better Experience & Pain Relief】Our keyboard arm rest is beneficial to alleviate the soreness caused by direct contact and friction between your arm and a hard desk surface, reducing the risk of wrist fatigue or carpal tunnel. The soft texture of memory foam can evenly distribute the pressure around your wrists and provide good support with just enough give.
- 【Helpful in Multiple Scenarios】Whether you're working, studying, writing, typing, gaming, this keyboard and mouse rest combo is an essential accessory to add comfort and support to your hands and wrists. It’s also a great gift for men, women, family, friend, coworker, gamer, teacher, etc.
Store the pending call and the continuation reference on the server rather than trusting values posted by the client. Before starting the next run, confirm that the reviewer can approve the tool, that the request has not expired, and that the stored arguments still satisfy current policy. If your application offers conditional review, implement that policy consistently for every entry point that can invoke the function.
How do I add approval gates in n8n?
Place a human-approval node or gate directly before the risky operation in the workflow. Configure the gate to pause execution and expose an approve or deny decision, then connect each outcome to an explicit branch. The workflow remains paused until the decision arrives.
Recommended Free Tools
Use separate gates for materially different effects instead of one generic approval at the end of a long workflow. Show the same target, arguments, requester, and expiration data in the approval interface, and ensure that a retry cannot execute the downstream node twice. n8n supplies the workflow pause, but identity checks, role authorization, redaction, and audit retention still belong in the surrounding application and deployment.
What security controls prevent replay or tampering?
Approval is safe only when the decision is bound to one immutable, server-owned proposal. Apply all of the following controls:
- Server-side state: keep serialized runs, arguments, and approval records in controlled storage. Do not let a browser submit a complete run snapshot.
- Authenticated identity: require a real application session, and recheck authorization for the specific tool, tenant, environment, and risk level.
- Opaque, single-use identifiers: use a decision identifier that maps to one pending record; consume it atomically and reject reuse.
- Argument immutability: load the stored call and never accept replacement arguments from the client. If editing is supported, store the edit as a controlled transformation and revalidate it.
- Integrity and ownership checks: verify that a serialized state belongs to the expected run and has not been altered before deserialization or resume.
- Expiry: reject decisions after the deadline, after policy changes that invalidate the request, or after the underlying resource has changed in a way your consistency rules do not allow.
- Fail-closed validation: malformed arguments, missing reviewer identity, unavailable policy data, or an unknown tool should stop execution rather than fall through.
- Output and UI hardening: redact secrets, escape model-controlled strings, and prevent HTML or script injection in review pages and logs.
- Least privilege: limit both the agent’s credentials and the reviewer’s approval scope; separate high-impact approvals from routine reads.
These controls address different threats. Authentication answers who is acting; authorization answers whether that person may act; atomic consumption answers whether the same approval can be replayed; argument immutability answers what will actually execute.
Can reviewers edit a proposed tool call?
Editing is appropriate only when the framework and policy explicitly support it. LangChain can allow approve, edit, reject, or respond outcomes; other integrations may provide only approve or reject. An edit must be limited to named fields—for example, changing a recipient or date but not switching the tool, tenant, environment, or permission scope.
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 minuteAfter an edit, parse and validate the new arguments, rerun risk and authorization checks, display the final call, and record both the original and edited values. Never let an edit turn a low-risk approval into an unreviewed high-impact operation.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How should timeout, retry, and failure behavior work?
Make the pending state explicit and design every failure path. A worker crash should leave a recoverable pending record, not trigger execution. A duplicate callback should return the current terminal status. A tool failure after approval should be reported as a failed execution, not as a new approval request unless the retry itself has a meaningful side effect.
Best Value
- 【No Drill Mounted】Cable management tray can be assembled in as little as 3 minutes or less-simply. Side clips easily clampe the edge of desk, pass through the holes on the sides, and secure excess cables with baskets. Desk cable management supports inward or outward installation to meet your needs in different scenarios, and provides you with great convenience when collecting and organizing wires.
- 【Rubber Pad Protects Desk】Wire organizer under desk built-in soft rubber pads on the clamp protect your desk from scratches, so you will not damage your furniture or office.
- 【Sturdy and Beautiful】Cord management under desk is made of high-quality carbon steel and holds about 15 lbs. The wire basket has 2 holes on each side of the tray to enable your cables to pass freely. Cable baskets are suitable for desk thickness use from 0.4" to 2.4".
- 【Space Saving & Keep Organized】Desk cable management tray holds all your power cords, outlet strips, USB hubs and computer transformers hidden from your feet, and the data cables will not enter the vacuum cleaner to keep your desk clean and tidy. Desk wire organizer is also suitable for use in the kitchen and outdoors.
- 【Reduce Safety Risks】Cable tray under desk keeps all plugs away from your kids, no more worry about tripping. Using 2 holes wire tray saves extra space, helps you easily cabling and wire protection. The mesh design is not easy to accumulate dust, but avoids safety accidents caused by messy wires.
- Timeout: expire the request and require a fresh proposal.
- Reviewer rejects: send a structured denial, stop, or request a revised plan according to the tool’s policy.
- Policy changes: invalidate pending requests affected by the new rule.
- Resource changes: recheck version, ownership, balance, or existence before executing if the operation depends on mutable state.
- Execution error: log the error and return a truthful failure result; never tell the model that the action succeeded.
How can I test an approval workflow before production?
Test the approval boundary as a security-critical state machine, not just as a user-interface feature. Use a controlled environment with non-production credentials and reversible fixtures.
- Submit valid and malformed arguments and confirm that invalid calls never execute.
- Attempt approval with an unauthenticated session, an unauthorized role, another tenant’s run, and an unrelated decision ID.
- Submit the same approval concurrently from two clients and verify that exactly one continuation runs.
- Change the stored arguments or policy version and confirm that tampering or stale approvals are rejected.
- Let a request expire, restart the worker, close the browser, and retry; none of these should approve silently.
- Exercise approve, edit, reject, and tool-error branches and inspect the resulting agent messages.
- Verify that secrets are redacted and that model-controlled strings cannot inject HTML or scripts.
- Check that the audit record contains the policy version, reviewer, decision, timestamps, final arguments, and tool result.
- For streaming, nested, and conversational runs, confirm that the original top-level session resumes with the correct state.
Primary framework documentation describes these implementation mechanisms and security requirements, but it does not establish a universal adoption rate, incident rate, latency figure, or success rate for human approval. Avoid presenting such statistics unless a named publisher and publication year support them.
Which approval framework should I choose?
Choose based on where you want policy, how much state management you need to own, and whether reviewers must edit calls.
| Framework | Approval mechanism | Decision choices | Pause and resume | Best fit |
|---|---|---|---|---|
| OpenAI Agents SDK | needsApproval; hosted MCP uses require_approval |
Approve or reject; sticky decisions can cover later calls | RunState serializes paused execution and resumes the original top-level run |
OpenAI agents with nested, streaming, or MCP tools |
| OpenAI API guidance | Application-owned human review paired with input, output, and tool guardrails | Approve or reject sensitive actions | Your application owns workflow state | Teams designing a policy boundary around side effects |
| LangChain/LangGraph | HITL middleware with configurable review policies | Approve, edit, reject, or respond | Persisted graph state enables pause and resume | Editable calls and framework-neutral orchestration |
| Microsoft Agent Framework | Function-tool approval_mode="always_require" |
Caller collects user input | Caller owns the continuation and starts the next run | Azure and Microsoft application stacks |
| n8n | Per-tool human-approval node or gate | Approve or deny | Workflow execution pauses until the decision | Low-code automation with clearly bounded risky tools |
Compare candidates on five questions: Is policy declared on the tool, in middleware, or in a workflow node? Can a reviewer edit arguments? How is paused state persisted? Are nested and streamed runs supported? Which authentication, authorization, and replay protections must your host application implement?
What is the practical design rule?
Make the side-effecting tool the last component that can say “yes.” The model can recommend an action, but only a server-side policy gate can authorize execution. Store the exact proposal, show its material consequences, require an authenticated and authorized decision, consume that decision once, and resume the same run with a truthful result. That arrangement scales from a single email-sending function to a multi-agent workflow without treating generated text as permission.
Frequently Asked Questions
Can I use automatic approval for low-risk agent actions?
Yes, if the tool is genuinely read-only or otherwise bounded and your policy documents that exception. Keep credentials, data scopes, and destinations restricted so that a supposedly low-risk call cannot be repurposed into a side effect.
Should an approval request include the agent’s full conversation?
Usually not. Show the triggering request and the evidence needed to understand the proposed effect, while keeping unrelated conversation and sensitive data server-side or behind a separately authorized view.
What if the underlying record changes while approval is waiting?
Treat the pending call as stale when your consistency rules detect a material change. Re-read the resource, invalidate the old request, and ask the agent to create a new proposal rather than applying an approval to an outdated target.
Can one approval cover several tool calls in a workflow?
Only when the policy explicitly defines the scope and the reviewer can see every resulting side effect. Otherwise approve each consequential call separately; bundling approvals can hide a later action that was not part of the original decision.
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.
