# AgentFide FAQ

Answer-first answers to common questions about AgentFide, the trust layer for the agent economy. Each entry leads with a direct answer, then adds detail.

## Start here

### What is AgentFide?
AgentFide is a neutral trust service for AI agents. Before an agent transacts, AgentFide verifies who it is dealing with, checks that the other side is authorized and reputable, and backs the transaction with recourse if something goes wrong, all in a single call.

### Explain AgentFide like I'm five.
Imagine robots that can spend money and make deals for you. Before a robot makes a deal, AgentFide checks if the other robot is real, allowed to make the deal, and known for being honest. If the deal goes bad, AgentFide helps you get your money back. It is like a trusted referee for robots that do business.

### What does the name AgentFide mean?
AgentFide combines "agent" with "fide," from the Latin word for faith and good faith, the root of bona fide, fidelity, and fiduciary. The name says what the product does: it confirms an agent is bona fide, or legitimate, and acting faithfully on behalf of whoever it represents.

### Who is AgentFide for?
AgentFide is for developers building AI agents, for platforms and marketplaces that need to gate the agents transacting on them, and for enterprises deploying agents that pay, buy, or hire. If an agent moves money or makes commitments, AgentFide is for whoever needs to trust that agent.

### Why do AI agents need a trust service at all?
Because agents can now transact in milliseconds, and there is no time for a human to check whether the agent on the other side is trustworthy. AgentFide answers that question automatically, so trust can keep up with the speed of agents.

### What problem does AgentFide solve?
It solves the trust gap in agent commerce. Agents can already pay and transact, but they cannot easily tell whether the other party is who it claims to be, is authorized, has a good record, and can be held to account. AgentFide answers all four in one call.

### Is AgentFide available today, and how do I get started?
Yes, AgentFide is live and self-serve. One plan at $49/month gives you the full assess API; you integrate in a single call and can have your agent verifying counterparties in minutes. No sales call is required to begin.

## How AgentFide works

### How does AgentFide work?
Your agent makes one assess call to AgentFide before it transacts. AgentFide runs four checks, identity, authority, reputation, and recourse, and returns a clear decision plus a credential the agent can present to the other side.

### What is an assess call?
The assess call is the single request your agent makes to AgentFide before a transaction. It carries who the agent is, who it acts for, and what it wants to do, and it returns an allow, review, or deny decision with the supporting details.

### What are the four checks AgentFide runs?
Identity confirms the agent and the operator behind it. Authority confirms the agent is acting within a valid, scoped mandate. Reputation scores the agent's track record. Recourse backs the transaction with escrow and the ability to undo a bad one.

### What does the identity check do?
It confirms the agent is real and tied to a known operator. AgentFide verifies the agent's signed credential, such as a signed A2A agent card, so you know exactly who you are dealing with rather than trusting a claim.

### What does the authority check do?
It confirms the agent is allowed to do what it is trying to do. AgentFide checks that the agent is acting within a valid, scoped mandate, with the right limits and an unexpired window, before any money moves.

### How does AgentFide score reputation?
AgentFide returns a transparent, explainable score built from an agent's real track record, things like whether it has paid, delivered, stayed in bounds, and avoided disputes. The score shows its factors rather than hiding behind a black box.

### What is recourse, and how does the undo work?
Recourse is your protection when a transaction goes wrong. AgentFide can hold funds in escrow and roll back a bad transaction, so a single mistake is recoverable rather than final. A score tells you the odds, recourse lets you act when the odds break.

### What does AgentFide return when my agent calls it?
A clear decision, allow, review, or deny, along with the identity result, the authority result, the reputation score and its factors, and whether recourse is available. It also returns a credential the agent can present to a counterparty.

### What standards does AgentFide use?
AgentFide is built on the standards agents already speak, including A2A signed agent cards for identity, the Model Context Protocol for integration, and AP2 for payment mandates. It also installs alongside the Stripe Agent Toolkit.

## Getting started

### How do I integrate AgentFide?
Add AgentFide as a tool your agent can call, then have it call assess before each transaction. It ships as an MCP server, a REST API, and SDKs, so most integrations are a single call and take minutes.

### Which agent frameworks does AgentFide support?
AgentFide works with the major agent frameworks, including the OpenAI Agents SDK, the Vercel AI SDK, LangChain, and CrewAI. Because it is also an MCP server, any agent that speaks the Model Context Protocol can use it.

### How do I use AgentFide as an MCP server?
Point your agent's MCP client at the AgentFide server and the assess tool becomes available to call directly. This is the easiest way to add AgentFide to any agent that speaks the Model Context Protocol.

### How long does integration take?
Minutes for a basic integration. The core is a single assess call before a transaction, so you can be verifying agents the same day you start.

### How do I get started with agent transactions safely?
Start by calling assess before any transaction your agent makes, then act on the decision: proceed on allow, pause on review, and stop on deny. Add recourse coverage for higher-value transactions so you are protected if something goes wrong.

### Do I need to change my payment setup to use AgentFide?
No. AgentFide runs alongside your existing payment setup. It decides whether to trust a transaction, it does not replace how you move money.

### Should my agent call AgentFide before every transaction?
Yes, that is the intended use. The assess call is fast and designed to sit in front of each transaction, so every deal your agent makes is checked before money moves.

## Trust and safety

### What happens when my agent buys from a bad agent?
If AgentFide flags the counterparty, your agent gets a review or deny decision and can stop before any money moves. If you took recourse coverage and a transaction still goes wrong, AgentFide can roll it back or recover the funds through escrow.

### How do I get my money back if a transaction goes wrong?
Through recourse. When a covered transaction fails or an agent misbehaves, AgentFide can undo the transaction or release escrowed funds back to you, according to the coverage policy.

### How does the undo or rollback work?
For covered transactions, AgentFide can reverse a bad transaction and return funds, rather than leaving you with a loss. This is the recourse layer, and it is what turns a reputation score into something you can act on.

### How are disputes handled?
Disputes resolve automatically by policy wherever possible, with no human step in the standard flow. Only true exceptions are escalated for manual handling.

### What does it mean if AgentFide returns deny or review?
Deny means the agent failed a check badly enough that you should not transact. Review means something needs a closer look before proceeding. Allow means the agent cleared the checks and you can proceed with confidence.

### Is my transaction data private?
Yes. Transaction details are private by default. What is public and verifiable is reputation, the score and badge, so a counterparty can check standing without seeing your business.

### Can an agent fake a good reputation?
AgentFide is built to resist that. Reputation is drawn from many independent sources rather than one, with provenance and weighting, so a single party cannot inflate its own agents or manufacture a record.

### Can I buy a better reputation score?
No. Reputation reflects real behavior and cannot be purchased. You can pay for verification, coverage, and a verified profile, but never for a higher score. That rule is what keeps AgentFide credible.

### How does AgentFide know an agent is legitimate?
It verifies the agent's signed credential and the operator behind it, rather than taking the agent's word. A verified signature ties the agent to a known, accountable operator.

### What if my own agent does something it should not?
The same recourse applies. If your agent acts outside its mandate or makes a bad transaction, AgentFide's checks can catch it and recourse can undo it, which limits the damage an autonomous agent can cause.

## AgentFide compared

### AgentFide vs building it yourself
Building it yourself means stitching together identity verification, a reputation system, escrow, and dispute handling, then sourcing enough data to make the scores meaningful. AgentFide gives you all of that in one call, with a cross-source reputation graph you could not easily build alone.

### AgentFide vs an identity or KYC provider
Identity providers tell you who someone is. AgentFide does identity too, then adds the parts that identity alone cannot answer: whether the agent is authorized, whether it has a good track record, and what your recourse is if it goes wrong.

### Why not just use A2A signed agent cards?
A signed agent card proves who an agent is, which is the identity piece, and AgentFide reads those cards. But a card says nothing about whether an agent is reputable, whether a specific action is authorized, or what happens when a deal goes bad. AgentFide adds those layers on top of the card.

### Does my payment provider already handle this?
Payment providers move money and handle payment fraud. They do not verify the agent on the other side, score its reputation across the ecosystem, or give you recourse on the transaction. AgentFide runs alongside your payment provider and answers the trust question it does not.

### AgentFide vs a traditional fraud or risk tool
Traditional fraud tools are built around human and card behavior. AgentFide is built for agents, scoring agent identity, authority, and reputation, and adding recourse designed for autonomous, machine-speed transactions.

### AgentFide vs an on-chain or blockchain reputation system
AgentFide gives you verifiable, portable reputation without putting your transactions on a public chain. Reputation stays checkable, transaction details stay private, and you avoid the cost and exposure of an on-chain ledger.

### Why not just trust the agent, or use human review?
Trusting blindly does not scale and invites loss, and human review cannot keep up with transactions that happen in milliseconds. AgentFide automates the trust decision so it is both fast and safe.

## Pricing

### How much does AgentFide cost?
AgentFide is one simple plan at $49/month for the full assess trust API. Enterprise features, higher volume, and recourse coverage are available on request.

### Is there a free tier?
There is no free tier; AgentFide is a single $49/month plan. The docs, the OpenAPI reference, and the SDKs are open, so you can evaluate the integration before you subscribe.

### How does pricing work as I scale?
Today it is one flat $49/month plan regardless of volume. Usage-based overage for very high volume is on the roadmap, and enterprise terms are available on request.

### What does recourse coverage cost?
Recourse coverage is a premium add-on priced by risk, so coverage for a trusted, high-reputation agent costs less than coverage for a risky one. It is rolling out; contact sales for early access.

### Do you charge to keep transactions private?
No. Privacy of transaction detail is the default, not an upsell. You are never charged to keep your transactions confidential.

## For platforms and enterprises

### Can a marketplace or platform embed AgentFide?
Yes. Platforms can use AgentFide as their built-in trust layer, gating the agents that transact on them and offering recourse, so the platform inherits trust without building it.

### What data can platforms get from AgentFide?
Platforms, rails, and underwriters can access reputation profiles, bulk scoring, and monitoring or alerts on the agents they care about, which helps them make and price their own trust decisions.

### Is AgentFide neutral, and who owns it?
AgentFide is neutral by design and is owned by no payment rail or marketplace. That neutrality is the point: any platform, rail, or agent can both feed it signals and rely on its decisions.

### Does AgentFide have enterprise features?
Yes. Enterprises can request enterprise features through a self-serve contact-sales step, which is the one place a human conversation begins. Everything else is self-serve.

### How does AgentFide stay neutral if it uses payment rails in the background?
AgentFide uses payment rails only as plumbing to move money, and it is rail-agnostic. The trust decision, the scoring, and the recourse logic are AgentFide's own and independent of any rail, so neutrality holds.

## About AgentFide

### Why is AgentFide neutral by design?
Because a trust service owned by one of the players it rates would not be credible, the way a credit bureau owned by a single lender would not be. Neutrality is what lets the whole ecosystem feed and rely on AgentFide.

### Who is behind AgentFide?
AgentFide was founded by Rob Steele. It is built as independent, neutral infrastructure for the agent economy rather than as the arm of any rail or marketplace.

### Is AgentFide built on open standards?
Yes. AgentFide builds on open agent standards, including A2A, the Model Context Protocol, and AP2, so it fits the tools and frameworks agents already use rather than locking you into a proprietary scheme.

### What is on the AgentFide roadmap?
The direction is a richer cross-source reputation graph, deeper recourse, and broad availability across agent frameworks and registries, so any transacting agent can reach AgentFide in one call.