Skip to content
AgentRadar

Methodology

How AgentRadar Rates AI Agents

How we discover agents and apply a consistent rubric to rate every listing in this directory

Our Approach

AgentRadar is an AI-assisted directory. We discover agents from public sources, build each profile from the agent’s own documentation and pricing pages, and score it with a fixed, weighted rubric. Ratings reflect public information — they are not the result of independent hands-on or lab testing. We never accept payment for ratings or rankings.

How we score

The Five Rating Dimensions

Each agent is scored 0–10 across five weighted dimensions, combined into the overall rating you see on every listing.

Ease of Use

20% weight

How quickly can a new user get value? Scored from onboarding flows, interface clarity, and documentation as described by the agent’s public materials.

Value for Money

25% weight

Is the price justified by what you get? Scored by comparing the published feature set and usage limits against the listed price, including free-tier generosity and paid-plan fairness.

Features

25% weight

How complete is the toolkit? Scored on the breadth and depth of capabilities relevant to the agent’s category, as listed in its public feature documentation.

Support

10% weight

When something breaks, can you get help? Scored from the agent’s public documentation, community channels, and stated response or SLA options.

Performance

20% weight

Does it deliver reliably? Scored from the agent’s stated speed, uptime, and output quality claims, plus any benchmark data the provider publishes.

The process

From Discovery to a Published Rating

How an agent goes from a public launch to a profile with a rating

1

1. Discover & Screen

We monitor launches, user submissions, and public signals (MCP Registry, Product Hunt, GitHub, and more). Each candidate is screened for legitimacy and active development before being profiled.

2

2. Profile & Score

Each agent’s features, pricing, and positioning are summarized from its public documentation and marketing materials, then scored against the rubric. We do not run independent hands-on tests.

3

3. Apply the Rubric

Dimension scores are assigned against the rubric and combined into the weighted overall rating you see on every listing.

4

4. Publish & Maintain

The profile is published with its sources. Profiles are refreshed on a regular cycle and whenever a major update, pricing change, or ownership shift is detected.

Our pledge

Independence & Accuracy

How we keep ratings fair

We may earn referral commissions when you sign up through affiliate links on this site. This never influences our ratings, rankings, or reviews. A higher commission never produces a higher score.

Our ratings are generated with AI assistance from public information using a fixed rubric. We do not accept paid placements, sponsored ratings, or "pay-to-rank" arrangements of any kind.

If we get something wrong, we want to know. Email us and we will investigate and correct verified inaccuracies promptly.

Browse all rated agents