Agent Access Levels.
Six levels, L0 to L5. Each says how far an AI agent gets in a product on its own: learn it, get in, use it, pay and leave.
Each level, and what it takes.
A level needs its own gate and every gate under it.
- L5
Autonomous
All of that, and it pays on its own for the work.
What it takes: It also pays on its own for the work: a payment we made and checked that buys credits or a plan for its account, or pays per call for at least 80% of the core. That is a payment on the product's agent payment route, such as x402 or MPP, or one purchase at its checkout on the owner's order with their discount code. A proven payment counts for 90 days. Or there is nothing to pay.
On a product card: Access, Use, Pay and Leave ticked.
- L4
Runs on its own
An agent does the core work and can stop paying on its own.
What it takes: It also does at least 80% of the core work and can stop paying on its own: it cancels any plan it started, or there was nothing to cancel.
On a product card: Access, Use and Leave ticked.
29 products at L4For example Attio
- Learn: Docs it can act on: yes: llms.txt, an OpenAPI and an MCP server.
- Access: Sign-up in a browser: yes.
- Use: 100% of the work: yes.
- Pay: Reaches the checkout: partly: Stopped at the card form.
- Leave: Leaves on its own: yes: Cancels on its own, deletes its account.
- L3
Does the work
An agent gets in on its own and does half the core work or more.
What it takes: It also does at least 50% of the core work, each task read back, on any way it got in.
On a product card: Access ticked, Use at least half.
13 products at L3For example Coda
- Learn: Docs with gaps: partly: An OpenAPI and an MCP server.
- Access: Sign-up in a browser: yes.
- Use: 70% of the work: partly.
- Pay: Card checkout: not reached.
- Leave: Nothing to cancel; its account stays: yes.
- L2
Gets in
An agent gets in on its own but does less than half the core work.
What it takes: The agent gets in on its own, by any way the product offers: its own sign-up, a key by API, an MCP sign-in, pay per call or no key at all. That holds in the latest drive and in at least half of the last three that could try that way.
On a product card: Access ticked.
10 products at L2For example Netlify
- Learn: Docs it can act on: yes: llms.txt, an OpenAPI and an MCP server.
- Access: Try first, claim later: yes: Can't sign up for one.
- Use: 40% of the work: no.
- Pay: Payment not reached: not reached.
- Leave: Leaves on its own: yes: Nothing to cancel, what it made goes after 1 hour.
- L1
Needs a person
Agents can't get in on their own.
What it takes: There is an API, CLI or MCP server, but the agent can't get in on its own: a person has to sign up, verify or make the key.
On a product card: Access not ticked: a person is in the way.
95 products at L1For example ActiveCampaign
- Learn: Docs it can act on: yes: llms.txt, an OpenAPI and an MCP server.
- Access: Can't sign up: no.
- Use: Work not reached: not reached.
- Pay: Payment not reached: not reached.
- Leave: Leaving not reached: not reached.
- L0
Read-only
Agents can read about it, but there is nothing to use.
What it takes: There is no API, CLI or MCP server, so there is nothing for an agent to use.
On a product card: Nothing to use.
4 products at L0For example Carrd
- Learn: Docs with gaps: partly.
- Access: Can't verify its account: no.
- Use: Work not reached: not reached.
- Pay: Payment not reached: not reached.
- Leave: Nothing to cancel; its account stays: yes.
How we measure a level.
Our agent uses the product itself, and code sets the level from what it did.
Our agent starts from the product's URL and works through nine checkpoints, from reading the docs to deleting its account. A task counts only when the agent reads the result back. The checkpoints
Any way in counts: its own sign-up, a key by API, an MCP sign-in, pay per call or no key at all. When a person has to help, the level stops at that step.
Bot checks don't always give the same answer. Getting in counts when the latest drive got in and at least half of the last three that could try that way did.
Levels and agent-readiness scores.
A readiness score reads what a site publishes for agents. A level comes from an agent using the product.
Reads what a site publishes for agents, such as robots.txt, a sitemap, markdown pages and an MCP server card. Cloudflare's agent-readiness scan and AgentReady work this way, and so does our Free check. None of them signs up, so none can show whether an agent gets a key or gets the work done.
Their checks next to oursOur agent signs up, gets a key, does the core work, pays and leaves. The level is how far it got on its own, and only a drive gives one.
A scan next to our drive, product by productThe Audited stamp.
A product at L5 whose owner had an Audit Drive carries the Audited stamp, with its month.
It never changes a level or a product's place in a list.
It has an end date, and it lapses when a later drive finds a lower level that isn't fixed.
See what an agent sees in yours.
A free check first. Then a drive proves what it can do.