# AgentDrives

> AgentDrives is an independent lab. We send AI agents to sign up for and use real software, and publish how far they get. For people learning how agents use software, for builders of agents, and for teams whose product agents use.

The directory is the lab's public record. Every product we drive is listed with its Agent Access Level, the date of the drive and where the agent stopped. Each drive takes our agent through nine checkpoints in five stages: find the product and its docs, get in (sign up, verify the account, get a key), use it (the first call and the core work), pay, and leave (cancel and delete).

Before your agent picks a tool, look it up:

- By task: GET https://agentdrives.com/api/find-tools?task=send+an+invoice gives the products that do it, best first, each with its level and where our agent stopped. A task none does answers covered false and, when we hold any, the nearest products in nearest_tools (match near, with what each lacks).
- One product: GET https://agentdrives.com/tools/<slug>.md. Find the slug with GET https://agentdrives.com/api/tools?query=<name or domain>.
- Over MCP, no key: https://agentdrives.com/api/mcp with find_tools_for_task and get_tool (listed at https://agentdrives.com/.well-known/mcp).
- As an Agent Skill: [check-agent-access](https://agentdrives.com/.well-known/agent-skills/check-agent-access/SKILL.md), listed at https://agentdrives.com/.well-known/agent-skills/index.json.
- Compare two or three products for a task: GET https://agentdrives.com/api/compare?tools=resend,postmark&task=send+an+email (MCP: compare_tools) gives each one's level, how far our agent got at each step, how to connect (MCP, API, CLI), whether we drove it or only listed it, the date and its page. Names we don't cover come back in not_covered.
- One product as JSON: GET https://agentdrives.com/tools/<slug>.json gives its status (driven or listed), level, each step, how to connect, the date and its page.
- Every API we offer: https://agentdrives.com/.well-known/api-catalog lists the REST API with its OpenAPI and the MCP server with its Server Cards (RFC 9727).
- Only a fetch tool? GET https://agentdrives.com/find?task=send+an+invoice answers the same search as a page, and https://agentdrives.com/find.md?task=send+an+invoice (or Accept: text/markdown) as Markdown.

Filters and fields:

- Filter find-tools with surface=api|mcp|cli|sdk, access=signup_without_human and pays_with=x402|mpp|card (the same on GET https://agentdrives.com/api/tools). Every result carries product_url, signup, docs_url, llms_txt_url, openapi_url, mcp_url, pays_with, free_tier and level (null when we don't know).
- What a product offers agents: GET https://agentdrives.com/api/tools/<slug>.json (MCP: get_tool), and each product in GET https://agentdrives.com/api/tools.json?via=llms and its pages, GET /api/tools and MCP search_tools, has agent_surfaces, for each of api, mcp, cli, sdk, docs and payment whether our drive tested it or the product documents it, even where our agent couldn't get in, and issues_found, how many issues the drive found. MCP get_agent_surfaces answers with just those, each surface with its source URL and the date it was seen.

Agent Access Levels: L0 Read-only (Agents can read about it, but there is nothing to use.); L1 Needs a person (Agents can't get in on their own.); L2 Gets in (An agent gets in on its own but does less than half the core work.); L3 Does the work (An agent gets in on its own and does half the core work or more.); L4 Runs on its own (An agent does the core work and can stop paying on its own.); L5 Autonomous (All of that, and it can pay on its own for the work.).

Part scores, 0-100, next to the level (never a total, never in the level; null = not measured): Holds up (Reliability and the crash test); Docs match (The docs checked against live behaviour). Security: reported privately to the owner, the same for every product.

Data: https://agentdrives.com/api/tools.json?via=llms (CC BY 4.0, what's free and what's licensed: https://agentdrives.com/data/licence?via=llms; 100 products a page: next gives the following page, full every product in one file). Every tool page has a markdown twin at /tools/<slug>.md and a JSON one at /tools/<slug>.json (driven or listed: status, level, steps, connect, as_of, page, licence). Filtered: GET https://agentdrives.com/api/tools?category=crm&min_level=4 (also access, e.g. mcp_signin: our agent signed in to the product's own MCP server, or key_by_api, browser_signup, pay_per_call; pays_with, e.g. x402; query; total says how many match; limit up to 100).

Every listing in one file: GET https://agentdrives.com/api/export (Team). An agent buys a plan itself: POST /api/orders {"product":"builder"} (or team), then pays it over HTTP 402; no domain to prove. Your own product's drive in full, every attempt with its evidence: GET https://agentdrives.com/api/detail/<slug> (MCP: drive_detail) with your key, once your claim on its domain is verified and with the Audit.

What changed, as Atom and JSON Feed: every product at https://agentdrives.com/feeds/changes.atom and https://agentdrives.com/feeds/changes.json; one product in the directory at /feeds/tools/<slug>.atom and .json; one category at /feeds/categories/<slug>.atom and .json; our rules at https://agentdrives.com/feeds/method.atom and .json. Each change names the two drives it comes from, with their ids and dates.

What an agent could pay for: GET https://agentdrives.com/api/payments (also product, from, to, limit): each drive's result at the pay step, dated, newest first: paid, refused, stopped or not_reached, with the reason. By day: https://agentdrives.com/feeds/payments.json (JSON Feed) and [Agent payments by day](https://agentdrives.com/data/payments?via=llms).

What changed this week in one call: GET https://agentdrives.com/api/changes?days=7 (also category, slug; MCP: what_changed): each product's level moves between drives with the cause, and the changes a second drive confirmed.
What changed at one product since you picked it: GET https://agentdrives.com/api/products/<slug>/changes (MCP: list_changes with slug): every dated change, with its source: level moves, surfaces that appeared or stopped answering, and the checkout's price and plan.

Filter the whole record: GET https://agentdrives.com/api/explore?category=crm&stop=signup (also area, capability, tag, level, checkpoint with outcome, vendor, from, to, release; format=csv for CSV): the products that match, each with its level, where our agent stopped and that stop's evidence line, with the set's checkpoint funnel and its source. Its filters' values: GET https://agentdrives.com/api/explore/options. The page: [Data explorer](https://agentdrives.com/explore).

Categories: GET https://agentdrives.com/api/categories.json lists the 12 areas and their categories, each with a stable id, a one-line definition and its products' counts. A category's products: https://agentdrives.com/categories/<slug>.json. The page: [Categories](https://agentdrives.com/categories?via=llms).

Find products by what an agent does: GET https://agentdrives.com/api/capabilities lists what an agent can do, by domain (e.g. Messaging: Send a transactional email), each with a stable id. GET https://agentdrives.com/api/find-tools?capability=<id or name> (MCP: find_tools_for_task with capability) answers with every product whose latest drive has a task of it: its level, what our agent got there, the drive date and the page section that shows it. In plain words instead: GET https://agentdrives.com/api/find-tools?task=send+an+invoice.

## For agents

Sign up in one call, with no person and no address: POST https://agentdrives.com/api/agents/signup {"agent":"your-name"} returns an API key at once and a claim link for your person (MCP tool: sign_up). Until a person claims the account or it pays for an Audit, it has 1 free check a day and an Audit, a Tool comparison or a data plan paid over HTTP 402.

A2A: our agent card is https://agentdrives.com/.well-known/agent-card.json (A2A 1.0, also 0.3). POST https://agentdrives.com/api/a2a with JSON-RPC SendMessage and a data part {"skill": "find_tools_for_task", "task": "send an invoice"}; the skills are find_tools_for_task, search_tools, get_tool (a product's listing; the verified owner's x-api-key adds its drive detail) and request_free_drive {"url": "…"}. No key needed; the same per-IP limit as the API.

Past a key's hourly limit a call answers 429 rate_limited with Retry-After and resets_at: wait that many seconds, the key still works. A 401 means the key is unknown, revoked or expired.

Everything else works without a browser too: sign in with an emailed code, create an API key, start a free check of any product, claim a domain, prove it is yours (an account email on the domain, which verifies at once unless anyone can get an address there (a mailbox provider, or agentdrives.com), a code mailed to any address at the domain, a DNS TXT record, a meta tag on the home page or a file at /.well-known/agentdrive-verify.txt), order drives, cancel, delete your account with your key (paid drives not run yet are cancelled and refunded; GET /api/me/delete lists them first). Orders for your own product need its domain verified first: a paid order before that gets 403 claim_not_verified, with every way to verify it. Getting started: https://agentdrives.com/getting-started.md (the calls in order) · Guide: https://agentdrives.com/data?via=llms#agents · OpenAPI: https://agentdrives.com/openapi.json?via=llms · MCP: https://agentdrives.com/mcp (send x-api-key for account tools); the directory tools alone, without a key: https://agentdrives.com/api/mcp.

Pay without a person: an Audit and its renewal, a Tool comparison (stack3, stack) and a data plan (builder, team) return payment.agent, at the listed price; a Tool comparison and a data plan need no domain proof, and an agent's own account can send "email" with the order for the receipt. POST https://agentdrives.com/api/orders/{id}/payment answers 402 for the order's price with MPP challenges (WWW-Authenticate: a card through a Stripe Shared Payment Token, or USDC on Tempo or Base) and x402 (PAYMENT-REQUIRED: USDC on Base); send the request again with the credential and the order starts (MCP tool: pay_order). Paid over HTTP 402, a data plan runs one month and doesn't renew by itself: pay again before it ends and the next month follows on. We email you 3 days before it ends. The 402's fix names the rails that are on. A business that wants our invoice sends {"buyer": {"vat_id": "DE123456789", "company": "Acme GmbH", "address": "1 Main St, 10115 Berlin"}} as the JSON body of both requests, or with the order. Every payment has a receipt at https://agentdrives.com/api/orders/{id}/receipt.

Pay per call: $0.01 a search, no rate limit, from prepaid credit or one call at a time. Credit: POST https://agentdrives.com/api/credits/topup {"amount_usd": 5} answers 402 with MPP (a card through a Stripe Shared Payment Token, or USDC on Base or Tempo) and x402 (USDC on Base); pay it and send the request again with the credential. Send an Idempotency-Key header with both requests: a retry never pays twice. $5 buys 500 calls, $20 buys 2,000, $50 buys 5,000. A balance holds at most $100 and lasts 12 months from the last top-up. It pays for calls only, not a report or a drive. Every top-up has a receipt. Without a key the top-up makes an agent account and returns its key (MCP tool: top_up). Then POST https://agentdrives.com/api/paid/find-tools {"task": "send an invoice"} with x-api-key answers at once and says what's left (credit in the body, X-Balance-Remaining); GET https://agentdrives.com/api/credits shows the balance (MCP tool: balance). One call at a time instead: without a key the call answers 402 with x402 and MPP, USDC only (a card's minimum is above the price). The same search is free at GET https://agentdrives.com/api/find-tools, rate-limited per IP.

## Docs

- [Getting started](https://agentdrives.com/getting-started.md): the calls an agent makes, in order, from finding a tool to signing up, verifying a domain, ordering, paying and leaving
- [API and MCP guide](https://agentdrives.com/data.md?via=llms): sign-up, keys, the MCP server, the data plans and webhooks
- [OpenAPI](https://agentdrives.com/openapi.json?via=llms): every route, its errors and the fix each error carries
- [API catalog](https://agentdrives.com/.well-known/api-catalog): our APIs in the standard RFC 9727 form: the REST API with its OpenAPI and guide, and the MCP server with its Server Cards
- [API changelog](https://agentdrives.com/data/api-changelog.md?via=llms): what changed in the API since 15 Oct 2026, by day: removed routes, retired products, new routes; the OpenAPI's info.version moves with each (JSON: https://agentdrives.com/data/api-changelog.json?via=llms)
- [Corrections](https://agentdrives.com/corrections.md?via=llms): how to report a wrong result, and every correction we made
- Every page has a Markdown twin at its address plus .md (the home: /index.md), with the same content as the page; send Accept: text/markdown to a page's own address and you get the twin

## Pricing

Your listing is free. The Audit gives you your fix list. Every product we drive is listed free with its level, the date and where an agent stops. The Audit adds your fix list with how to make each fix, your full profile for 12 months on our site and in our API and MCP, and new drives after your fixes. The whole offer as JSON: https://agentdrives.com/pricing.json?via=llms. Order over the API (POST /api/orders, product ids below), the MCP order tool, or a person in the dashboard: https://agentdrives.com/app/orders/new?plan=<product id>.

- Free check ("product": "free"): $0, private. See what an agent sees. What an agent finds without a browser: docs, API, MCP, llms.txt; Any wall in its way, and how many things to fix; How the products we drove in your category did. One per domain every 90 days. In the directory: nothing, it stays private.
- Audit ("product": "audit"): $499 the first year. We drive your whole product, and again as you fix. When an agent can use it fully with no person, a certificate, valid 12 months. Our agent signs up like a new customer; Every API route, MCP tool and CLI command you document, checked; A security review, for you only; A real purchase in your checkout, with your code; Your fix list: every fix and how to make it; We drive you again after your fixes; Your full profile for 12 months, on our site and in our API and MCP; When an agent can use it fully on its own, the Audited badge and a certificate, valid 12 months. Then $199 a year. Paid by card, it renews by itself a month before each year ends. We email you 30 days before; cancel with one click. In the directory: your full profile for 12 months.
- A free check's report is usually ready 1 to 5 minutes after the order: the order's status turns done (GET /api/orders/{id}, MCP my_orders), the report is at GET /api/orders/{id}/report (MCP order_report), and an account with an email address gets the result by mail at once.
- The free check needs no domain proof: any signed-in account (offers_ok: true only if you want the monthly report and offers); one per domain every 90 days (within them the call answers with the domain's latest check and next_free_at) and 3 a day per account. An Audit needs a verified claim on the product's domain (403 claim_not_verified otherwise); once it is proved, GET https://agentdrives.com/api/me/reports/{slug} shows what the last drive found, counted, and the level its fixes can reach. Its security review is always included. It takes target: the level you aim for, 1 to 5 (Your goal shows on your dashboard. The level you get is the one the drives find.), and the owner's paid plan at cost (plan). After your fixes, POST /api/orders/{id}/followup drives it again, one at a time and at most once a day, within the Audit year (12 months from the report); POST /api/orders/{id}/finish ends it early at L5 (open fixes go on the certificate as Not done). The certificate is only for a product an agent uses fully on its own, with no person at any step (L5). You pay for the audit, never for a level.
- Audit renewal ("product": "renewal"): $199 a year. Another year of your full profile: a new Audit of your product, and new drives after your fixes. Paid by card, it renews by itself a month before each year ends. We email you 30 days before. Paid another way, you can renew from a month before your year ends. Findings, the Full report and their upgrades are retired (product drive, findings, quick or report, and upgrade_from, answer 400 product_retired).
- Watch: Every week our agent signs up and uses your product again, and tells you the day that breaks and the day it's fixed. Only you see the runs. Watch 3: $79 a month or $790 a year, up to 3 products; Watch 10: $199 a month, up to 10 products. For products you've proved are yours. Cancel any time: it runs to the end of the period you paid. Start it signed in: POST /api/watch {"plan": "watch3"} (or watch3_year, watch10), or https://agentdrives.com/app/billing#watch; not through /api/orders.
- Again: After a fix, our agent drives it again for free, up to twice a year.
- Re-drives: A listing never lapses because nobody pays. It shows the date of its drive. Fixed something? Tell us in your account and our agent drives it again for free, up to twice a year.
- How the directory works: A product is in the directory because an agent drove it. We pick products ourselves, and anyone can suggest one for free; an owner can also order a drive of their own product. Both drives are the same, listed at whatever level they find, and an owner's order is labelled. An owner's order says "Owner is an AgentDrives customer" on its listing and in the API. Suggest a product: POST /api/suggest {url, email?}.
- Tool comparison · 3 tools: $490 one time. For a team choosing tools for its agents; We drive three tools and put them side by side; Which ones an agent can sign up for, get a key and use; Where an agent stops in each.
- Tool comparison · 5 tools: $650 one time · +$130 a tool after five, up to ten. For a team choosing tools for its agents; We drive five tools and put them side by side; Which ones an agent can sign up for, get a key and use; What each one needs from you to work for your agents; A walk-through of the result.
- Fair use (Audit): One Audit covers one product: one sign-up, and the API, MCP server and CLI that come with it. We read and check every documented operation, whatever the size, and run every one that is safe on our own account within one drive of up to three hours. On very large products we run all reads, everything behind your core tasks and at least one call per resource group, and the report says exactly what was sampled. Your next drive picks up the rest. After your fixes we drive it again when you ask, once a day at most, within the year.
- Badges: Driven (a drive, with its level and date), and the Audited mark of an Audit for a product an agent uses fully with no person (L5).
- Not an owner: Tool comparison (above) for teams choosing tools for their agents; data plans for agent builders at https://agentdrives.com/data?via=llms.
- Crash test (the resilience part): every drive tests how the product handles an agent's mistakes once it is in, on our agent's own account: a few malformed requests, a wrong key of its own, a retried create, 10 to 20 reads of one endpoint (stopping at the first 429) and each core feature run three times.
- Prices are in US dollars; buyers in the EEA pay the same number in euros. Every payment comes with a receipt or our invoice. A report that needs a new drive comes within 48 hours of payment. Details: https://agentdrives.com/pricing?via=llms

## Tools

Every product we drove, 193 in all.

Listed, not driven: 462 more products whose agent surfaces we read in their public docs. No level until a drive: https://agentdrives.com/api/listed.json

One line per product (its level, category and the drive's first sentence), a file per category; every drive in full: https://agentdrives.com/llms-full.txt?via=llms.

- [Browser infrastructure](https://agentdrives.com/llms/browser-infrastructure.txt?via=llms): 4 products
- [CRM](https://agentdrives.com/llms/crm.txt?via=llms): 12 products
- [Customer support](https://agentdrives.com/llms/customer-support.txt?via=llms): 10 products
- [Databases](https://agentdrives.com/llms/databases.txt?via=llms): 14 products
- [Email infrastructure](https://agentdrives.com/llms/email-infrastructure.txt?via=llms): 8 products
- [Email marketing](https://agentdrives.com/llms/email-marketing.txt?via=llms): 9 products
- [File storage](https://agentdrives.com/llms/file-storage.txt?via=llms): 6 products
- [Forms](https://agentdrives.com/llms/forms.txt?via=llms): 7 products
- [Invoicing](https://agentdrives.com/llms/invoicing.txt?via=llms): 5 products
- [Product analytics](https://agentdrives.com/llms/product-analytics.txt?via=llms): 4 products
- [Project management](https://agentdrives.com/llms/project-management.txt?via=llms): 15 products
- [Scheduling](https://agentdrives.com/llms/scheduling.txt?via=llms): 6 products
- [Search APIs](https://agentdrives.com/llms/search-apis.txt?via=llms): 6 products
- [Social media scheduling](https://agentdrives.com/llms/social-media-scheduling.txt?via=llms): 5 products
- [Task management](https://agentdrives.com/llms/task-management.txt?via=llms): 4 products
- [Web analytics](https://agentdrives.com/llms/web-analytics.txt?via=llms): 4 products
- [Web scraping](https://agentdrives.com/llms/web-scraping.txt?via=llms): 4 products
- [Docs and workspace](https://agentdrives.com/llms/workspace.txt?via=llms): 5 products
- [E-signature](https://agentdrives.com/llms/e-signature.txt?via=llms): 4 products
- [Payments](https://agentdrives.com/llms/payments.txt?via=llms): 7 products
- [Website builders](https://agentdrives.com/llms/website-builders.txt?via=llms): 5 products
- [E-commerce](https://agentdrives.com/llms/ecommerce.txt?via=llms): 4 products
- [Team chat](https://agentdrives.com/llms/team-chat.txt?via=llms): 7 products
- [Design](https://agentdrives.com/llms/design.txt?via=llms): 5 products
- [Hosting and deploys](https://agentdrives.com/llms/hosting.txt?via=llms): 6 products
- [Automation](https://agentdrives.com/llms/automation.txt?via=llms): 4 products
- [Vector databases](https://agentdrives.com/llms/vector-databases.txt?via=llms): 5 products
- [Data warehouses](https://agentdrives.com/llms/data-warehouses.txt?via=llms): 6 products
- [Source control](https://agentdrives.com/llms/source-control.txt?via=llms): 5 products
- [Push notifications](https://agentdrives.com/llms/push-notifications.txt?via=llms): 1 product
- [Billing and subscriptions](https://agentdrives.com/llms/billing.txt?via=llms): 1 product
- [Accounting](https://agentdrives.com/llms/accounting.txt?via=llms): 1 product
- [CI/CD](https://agentdrives.com/llms/ci-cd.txt?via=llms): 1 product
- [API testing and mocking](https://agentdrives.com/llms/api-testing.txt?via=llms): 3 products

## Categories

- [Browser infrastructure for AI agents](https://agentdrives.com/categories/browser-infrastructure.md?via=llms): 4 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/browser-infrastructure.json?via=llms
- [CRM for AI agents](https://agentdrives.com/categories/crm.md?via=llms): 12 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/crm.json?via=llms
- [Customer support for AI agents](https://agentdrives.com/categories/customer-support.md?via=llms): 10 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/customer-support.json?via=llms
- [Databases for AI agents](https://agentdrives.com/categories/databases.md?via=llms): 14 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/databases.json?via=llms
- [Email infrastructure for AI agents](https://agentdrives.com/categories/email-infrastructure.md?via=llms): 8 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/email-infrastructure.json?via=llms
- [Email marketing for AI agents](https://agentdrives.com/categories/email-marketing.md?via=llms): 9 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/email-marketing.json?via=llms
- [File storage for AI agents](https://agentdrives.com/categories/file-storage.md?via=llms): 6 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/file-storage.json?via=llms
- [Forms for AI agents](https://agentdrives.com/categories/forms.md?via=llms): 7 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/forms.json?via=llms
- [Invoicing for AI agents](https://agentdrives.com/categories/invoicing.md?via=llms): 5 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/invoicing.json?via=llms
- [Product analytics for AI agents](https://agentdrives.com/categories/product-analytics.md?via=llms): 4 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/product-analytics.json?via=llms
- [Project management for AI agents](https://agentdrives.com/categories/project-management.md?via=llms): 15 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/project-management.json?via=llms
- [Scheduling for AI agents](https://agentdrives.com/categories/scheduling.md?via=llms): 6 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/scheduling.json?via=llms
- [Search APIs for AI agents](https://agentdrives.com/categories/search-apis.md?via=llms): 6 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/search-apis.json?via=llms
- [Social media scheduling for AI agents](https://agentdrives.com/categories/social-media-scheduling.md?via=llms): 5 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/social-media-scheduling.json?via=llms
- [Task management for AI agents](https://agentdrives.com/categories/task-management.md?via=llms): 4 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/task-management.json?via=llms
- [Web analytics for AI agents](https://agentdrives.com/categories/web-analytics.md?via=llms): 4 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/web-analytics.json?via=llms
- [Web scraping for AI agents](https://agentdrives.com/categories/web-scraping.md?via=llms): 4 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/web-scraping.json?via=llms
- [Docs and workspace for AI agents](https://agentdrives.com/categories/workspace.md?via=llms): 5 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/workspace.json?via=llms
- [E-signature for AI agents](https://agentdrives.com/categories/e-signature.md?via=llms): 4 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/e-signature.json?via=llms
- [Payments for AI agents](https://agentdrives.com/categories/payments.md?via=llms): 7 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/payments.json?via=llms
- [Website builders for AI agents](https://agentdrives.com/categories/website-builders.md?via=llms): 5 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/website-builders.json?via=llms
- [E-commerce for AI agents](https://agentdrives.com/categories/ecommerce.md?via=llms): 4 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/ecommerce.json?via=llms
- [Team chat for AI agents](https://agentdrives.com/categories/team-chat.md?via=llms): 7 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/team-chat.json?via=llms
- [Design for AI agents](https://agentdrives.com/categories/design.md?via=llms): 5 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/design.json?via=llms
- [Hosting and deploys for AI agents](https://agentdrives.com/categories/hosting.md?via=llms): 6 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/hosting.json?via=llms
- [Automation for AI agents](https://agentdrives.com/categories/automation.md?via=llms): 4 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/automation.json?via=llms
- [Vector databases for AI agents](https://agentdrives.com/categories/vector-databases.md?via=llms): 5 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/vector-databases.json?via=llms
- [Data warehouses for AI agents](https://agentdrives.com/categories/data-warehouses.md?via=llms): 6 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/data-warehouses.json?via=llms
- [Source control for AI agents](https://agentdrives.com/categories/source-control.md?via=llms): 5 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/source-control.json?via=llms
- [Push notifications for AI agents](https://agentdrives.com/categories/push-notifications.md?via=llms): 1 product by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/push-notifications.json?via=llms
- [Billing and subscriptions for AI agents](https://agentdrives.com/categories/billing.md?via=llms): 1 product by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/billing.json?via=llms
- [Accounting for AI agents](https://agentdrives.com/categories/accounting.md?via=llms): 1 product by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/accounting.json?via=llms
- [CI/CD for AI agents](https://agentdrives.com/categories/ci-cd.md?via=llms): 1 product by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/ci-cd.json?via=llms
- [API testing and mocking for AI agents](https://agentdrives.com/categories/api-testing.md?via=llms): 3 products by level, where agents drop out, what an agent can do in each, the walls. Data: https://agentdrives.com/categories/api-testing.json?via=llms

## Articles

Every number in an article is a fact computed from the drives (https://agentdrives.com/data/facts?via=llms); each article has a markdown twin.

- [CLI for AI agents: 11 of 14 databases have one](https://agentdrives.com/research/which-kinds-of-software-give-ai-agents-a-cli-or-mcp-server.md?via=llms): Data story, 2026-10-31. An agent working in forms, invoicing or file storage has the API or the MCP server to work through, where those are listed. Numbers: 193 products, 1–31 Oct 2026.
- [Can AI agents pay for software? 28 products offer a way](https://agentdrives.com/research/can-ai-agents-pay-for-software.md?via=llms): Data story, 2026-10-31. An owner can sell to agents by answering an unpaid call with an MPP payment request that Stripe settles to its balance. Numbers: 193 products, 1–31 Oct 2026.
- [6 of 193 products let an AI agent finish the job alone](https://agentdrives.com/research/software-isnt-ready-for-agent-customers.md?via=llms): Data story, 2026-10-31. Docs are the easy part for an AI agent. Sign-up and payment are where a person still has to step in. Numbers: 193 products, 1–31 Oct 2026.
- [Bot checks on SaaS checkouts: where AI agents stop](https://agentdrives.com/guides/bot-checks-on-saas-checkouts-where-ai-agents.md?via=llms): Guide, 2026-10-31. 42 of 59 SaaS checkouts our agent reached carried a bot check; under 5 made it solve one. 80 of 139 stopped earlier. What to fix so an agent can buy.
- [Does an MCP server mean an AI agent can use your product?](https://agentdrives.com/blog/why-we-drive-software-with-agents.md?via=llms): Blog, 2026-10-31. Neon, Cloudinary and Apify handed our agent a working credential through their own routes, with no sign-up form in its way. Numbers: 191 products, 1–29 Oct 2026.
- [Which bot checks stopped our AI agent? 109 of 191 products](https://agentdrives.com/research/which-bot-checks-stop-ai-agents.md?via=llms): Data story, 2026-10-30. Some bot checks put a challenge on the page that a person could clear; others refused our agent with nothing shown, so there was nothing to hand over. Numbers: 191 products, 1–29 Oct 2026.
- [Can an AI agent delete its account? 59 products](https://agentdrives.com/guides/can-an-ai-agent-delete-its-own-account.md?via=llms): Guide, 2026-10-30. Our agent deleted its own account on 40 of 59 products (1–29 Oct 2026), only 6 over the API. What blocked it and what to change.
- [How to let AI agents sign up for your product](https://agentdrives.com/guides/let-ai-agents-sign-up.md?via=llms): Guide, 2026-10-30. AgentMail and Cloudinary hand an agent a working key in one call; copy their flow with a short onboarding API of your own. Numbers: 191 products, 1–29 Oct 2026.
- [Best databases for AI agents: 13 driven, ranked](https://agentdrives.com/research/best-databases-for-ai-agents.md?via=llms): Data story, 2026-10-29. We drove 13 databases with a real agent, 3 to 27 Oct 2026: the level each reached, how it got in, the core work done and where it stopped. Numbers: 13 products, 3–27 Oct 2026.
- [Can AI agents sign up for software? We drove 185](https://agentdrives.com/research/can-ai-agents-sign-up.md?via=llms): Data story, 2026-10-29. Where sign-up stopped our agent first, a bot check was the reason at 91 of 109 products and a required card field at Fathom Analytics. Numbers: 185 products, 1–29 Oct 2026.
- [Email verification for AI agents: which mail works, how fast](https://agentdrives.com/guides/email-verification-for-ai-agents.md?via=llms): Guide, 2026-10-29. Which sign-up verification mail our AI agent finished without a browser, how fast the mail reached its inbox, and what to change in your own flow.
- [Only 17 of 55 API keys our agent got came without a browser](https://agentdrives.com/research/api-keys-without-a-browser.md?via=llms): Data story, 2026-10-28. 61 of 66 first API or MCP calls our agent reached needed no browser. The key was where it needed one: it copied 36 keys from a dashboard. Numbers: 168 products, 1–27 Oct 2026.
- [64% of SaaS pricing pages give an AI agent a readable price](https://agentdrives.com/guides/saas-pricing-pages-ai-agents-can-read.md?via=llms): Guide, 2026-10-28. We requested /pricing at 106 SaaS products with no browser, 1–28 Oct 2026: 68 gave a price an agent can quote. What the rest get wrong and how to fix it.
- [98 of 101 SaaS products offer sign-up with email alone](https://agentdrives.com/guides/saas-products-offer-email-only-sign-up.md?via=llms): Guide, 2026-10-28. We drove sign-up at 101 products (Oct 2026): 98 accept an email address alone, social-only is under 5, and a phone check stopped our agent at 5 of 101.
- [Tell an AI agent a feature needs a paid plan](https://agentdrives.com/guides/tell-ai-agents-a-feature-needs-a-paid-plan.md?via=llms): Guide, 2026-10-28. How to answer an AI agent at a paid-plan wall: 402 or 403, a Problem Details body with the required plan, and an upgrade link a person can open.
- [Write API errors an AI agent can act on](https://agentdrives.com/guides/api-errors-agents-can-act-on.md?via=llms): Guide, 2026-10-27. How to shape API errors an AI agent can act on: status codes, Problem Details, Retry-After and request ids, with a baseline from live product APIs.
- [Resend publishes an MCP Server Card and an AI Catalog](https://agentdrives.com/notes/2026-10-27-resend-publishes-an-ai-catalog-and-an-mcp-server.md?via=llms): Note, 2026-10-27. Clients that support the new files can find Resend's MCP server from its domain. Our agent got into 51 of 134 products with an MCP server on its own. Numbers: 155 products, 1–26 Oct 2026.
- [Do signed AI agents get past bot checks?](https://agentdrives.com/research/who-gets-in-servers-signed-agents-browsers.md?via=llms): Data story, 2026-10-24. A challenge on the sign-up form stopped our signed agent at 38 of 94 products where a bot check stopped it. Numbers: 151 products, 1–24 Oct 2026.
- [AWS now lists DataDome and F5 for Web Bot Auth](https://agentdrives.com/notes/2026-10-23-aws-adds-datadome-and-f5-to-agentcore-browser-s.md?via=llms): Note, 2026-10-24. AWS lists DataDome and F5 for its signed agents. Cloudflare stopped our signed, unverified agent at 34 of 34 products where it met it. Numbers: 148 products, 1–22 Oct 2026.
- [Cloudflare starts scoring how its listed bots behave](https://agentdrives.com/notes/2026-10-19-cloudflare-starts-scoring-how-listed-bots-behave.md?via=llms): Note, 2026-10-24. Cloudflare says it rates Verified bots internally on robots.txt and HTTP 429s. It stopped our signed agent at 33 of 33 products where it met it. Numbers: 140 products, 1–18 Oct 2026.
- [How far does an AI agent get in SaaS on its own? 148 tested](https://agentdrives.com/reports/how-far-a-real-agent-got-2026-10.md?via=llms): Report, 2026-10-23. No product reached Autonomous: 27 run on their own, and five took an agent payment with no card. Numbers: 148 products, 1–22 Oct 2026.
- [Stripe's SPT docs show a Resolve API for other processors](https://agentdrives.com/notes/2026-10-16-stripe-spt-docs-add-a-resolve-api-for-third-party.md?via=llms): Note, 2026-10-16. What a seller using another processor needs to take an agent's SPT, in preview. We proved a native agent payment at 5 of 140. Numbers: 140 products, 1–16 Oct 2026.

## Standards

Eight open standards a product can publish or follow for AI agents: what each is, how many of the products we drove use it, which ones, and how to add it. The list as Markdown: https://agentdrives.com/standards.md

- [MCP](https://agentdrives.com/standards/mcp.md): A server that turns a product's actions into tools an agent can call.
- [llms.txt](https://agentdrives.com/standards/llms-txt.md): A Markdown map of a product's docs, written for language models.
- [OpenAPI](https://agentdrives.com/standards/openapi.md): A machine-readable description of every route of an API.
- [Web Bot Auth](https://agentdrives.com/standards/web-bot-auth.md): An agent signs each request; a site's bot check can verify who sent it.
- [Signed agents](https://agentdrives.com/standards/signed-agents.md): Agents a bot-check vendor has verified, which a site can let through.
- [robots.txt for AI agents](https://agentdrives.com/standards/robots-txt-ai.md): The file that tells crawlers and agents which pages to leave alone.
- [x402](https://agentdrives.com/standards/x402.md): A 402 answer an agent pays in a stablecoin, request by request.
- [MPP](https://agentdrives.com/standards/mpp.md): A 402 payment challenge an agent pays by card or stablecoin.

## Guides

For owners: how to make a product work for agents, and one fix guide per wall our agent meets, with the steps and a check to run. The list as Markdown: https://agentdrives.com/guides.md?via=llms

- [Make your SaaS agent-ready: the whole path, step by step](https://agentdrives.com/guides/make-your-saas-agent-ready.md?via=llms): Docs, sign-up, the key, MCP, paying and leaving: what an agent needs at each step, with the fix for each.
- [An MCP server is not a way in for AI agents](https://agentdrives.com/guides/an-mcp-server-is-not-a-way-in.md?via=llms): Most products have an MCP server. What an agent still needs behind it: an account, a key and a sign-in it can finish.
- [How AI agents pay for software: x402, MPP and checkouts](https://agentdrives.com/guides/how-ai-agents-pay-for-software.md?via=llms): Card checkouts, x402 and MPP: what an agent can pay with today, and what most products ask instead.
- [llms.txt, OpenAPI and MCP alone don't let AI agents in](https://agentdrives.com/guides/agent-ready-files-dont-open-the-door.md?via=llms): Most products publish llms.txt, OpenAPI or MCP. Most still keep agents out. What the files do, and what they can't.
- [Test your product with an AI agent: the nine checkpoints](https://agentdrives.com/guides/test-your-product-with-an-ai-agent.md?via=llms): The nine checkpoints of a drive, with the commands to try each one from a terminal.
- [Let signed AI agents past your bot check](https://agentdrives.com/guides/let-signed-agents-past-your-bot-check.md?via=llms): The setting for Cloudflare, Vercel, Akamai, DataDome and HUMAN, and what to do about a check on the form.
- [Let AI agents sign up and get an API key](https://agentdrives.com/guides/let-agents-sign-up-and-get-a-key.md?via=llms): An API call that signs up and returns a limited key, and a claim step for the person.
- [Let AI agents sign up without a phone number](https://agentdrives.com/guides/sign-up-without-a-phone-number.md?via=llms): No phone on the first step: an email code, a person who approves the agent, limits until they do.
- [MCP sign-in an AI agent can finish: OAuth and CIMD](https://agentdrives.com/guides/mcp-sign-in-agents-can-finish.md?via=llms): OAuth metadata, CIMD or DCR, and a way to approve without a browser on the agent's side.
- [Let AI agents pay you per call: x402 and MPP](https://agentdrives.com/guides/let-agents-pay-per-call.md?via=llms): x402 or MPP on a paid endpoint, your prices where agents read them, and plans an agent can buy by API or MCP.
- [Let AI agents delete the account they opened](https://agentdrives.com/guides/let-agents-delete-their-account.md?via=llms): A delete call for the account, a clear answer when it's done, and the same in the settings for people.

## Research

- [Sign-up is the wall](https://agentdrives.com/findings/the-front-door.md?via=llms): Our agent got into 73 of 193 products on its own. At the other 120, it needed a person or was stopped. Drives up to 31 Oct 2026.
- [MCP servers everywhere, agents nowhere](https://agentdrives.com/findings/mcp-without-access.md?via=llms): 100 of the 167 products with an MCP server keep an agent out: it can't get in on its own to use the server.
- [Agents can pay, but only next to the product](https://agentdrives.com/findings/agents-pay-at-the-side-door.md?via=llms): 1 of the 5 products that take agent-native payment don't take it for the product you sign up for.
- [Agents can't leave](https://agentdrives.com/findings/no-way-out.md?via=llms): 20 of the 60 products where our agent got as far as deleting its account gave it no way to do it on its own.
- [Bot checks, provider by provider](https://agentdrives.com/research/bot-checks.md?via=llms): per vendor, how often it stopped our signed agent and the setting that lets signed agents through
- [State of agent access](https://agentdrives.com/research/state-of-agent-access.md?via=llms): our monthly report on agent access, every edition with its data
- [State of agent access, October 2026](https://agentdrives.com/research/state-of-agent-access-2026-10.md?via=llms): how far our agent got in the products we drove, and what it met at the 5,000 most-linked software sites, 12–13 Oct 2026. Data: https://agentdrives.com/research/state-of-agent-access-2026-10.json?via=llms

## Pages

Every page as Markdown, the same content as the page:

- [The directory](https://agentdrives.com/tools.md?via=llms): every product we drove, by level
- [Categories](https://agentdrives.com/categories.md?via=llms)
- [Research](https://agentdrives.com/research.md?via=llms)
- [Rankings](https://agentdrives.com/rankings.md)
- [Glossary](https://agentdrives.com/glossary.md)
- [Awards](https://agentdrives.com/awards.md)
- [How we test](https://agentdrives.com/method.md?via=llms)
- [The levels](https://agentdrives.com/method/levels.md?via=llms)
- [The rules](https://agentdrives.com/method/rules.md?via=llms)
- [What changed in the method](https://agentdrives.com/method/changelog.md?via=llms)
- [The AgentDrives Standard](https://agentdrives.com/standard.md): every criterion a product meets so an AI agent can use it on its own, with what to fix (JSON: https://agentdrives.com/standard.json)
- [Data, API and MCP](https://agentdrives.com/data.md?via=llms)
- [Facts: every number with its evidence](https://agentdrives.com/data/facts.md?via=llms)
- [Bot checks per product](https://agentdrives.com/data/bot-checks.md?via=llms): which stopped our signed agent and which let it in, updated daily (JSON: https://agentdrives.com/api/bot-checks, CSV: https://agentdrives.com/api/bot-checks.csv)
- [Data licence](https://agentdrives.com/data/licence.md?via=llms)
- [Pricing](https://agentdrives.com/pricing.md?via=llms): the offer as JSON: https://agentdrives.com/pricing.json?via=llms
- [The Audit](https://agentdrives.com/audit.md)
- [Example report](https://agentdrives.com/example-report.md?via=llms)
- [Free check](https://agentdrives.com/scan.md?via=llms)
- [Tool comparison](https://agentdrives.com/stack-check.md)
- [Our own drive](https://agentdrives.com/our-own-drive.md)
- [Corrections](https://agentdrives.com/corrections.md?via=llms)
- [About](https://agentdrives.com/about.md?via=llms)
- [Contact](https://agentdrives.com/contact.md?via=llms)
- [Security](https://agentdrives.com/security.md?via=llms)
- [Press kit](https://agentdrives.com/brand.md)
- [Listing policy](https://agentdrives.com/legal/listing-policy.md?via=llms)
- [Privacy](https://agentdrives.com/legal/privacy.md?via=llms)
- [Terms](https://agentdrives.com/legal/terms.md?via=llms)