# Fix list: Paid From Anchor Terminal's listing at https://www.anchorterminal.com/tools/paid, the October 2026 research run, assessed 9 October 2026. Grade C, 55.1 out of 100. This is everything the published grade says the listing lacks, the biggest possible gain to the total first. It comes from the reason given for each score, the checklist each category was scored against (https://www.anchorterminal.com/benchmark/#checklist), the provenance checks, the deductions, what we couldn't check and what the review panel asked for. A fix counts at the next check, once it's public. For a coding agent working on Paid: work through the items below in the product, its docs and its public pages. Each category gives the reason for its score, with the points each checklist item earned, and the checklist itself, so the gap is the items that earned less than their points. Change the product, not the wording, and keep a note of what you changed and where it's published. ## 1. Reliability, 35 out of 100, up to 13 more on the total Why it scored 35: Hosted lines. No status page was found in the pricing page footer, the docs index or the pages read, and trust.paid.ai was not read (0). With no history to read, the incident line takes 5. Rate limits are described as a token bucket per organisation on signal ingestion, counted per signal, with a 500-signal cap a request, but no bucket size or refill rate is published and other endpoints are not yet limited (7 of 15). A 429 carries `Retry-After`, the docs ask for exponential backoff with jitter, a per-signal `idempotencyKey` makes batch retries safe and the SDKs retry twice by default. No idempotency key was found on orders or payments (13 of 15). The agreement promises commercially reasonable efforts at 24-hour availability and 48 hours' notice of planned downtime, with no percentage or credit (0). The v2 API is the default version with no beta label, and custom views are marked experimental (10). The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability): Hosted APIs, MCP servers, models and platforms. - 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own). - 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so. - 15, rate limits documented with numbers. - 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved. - 10, an SLA published for any paid tier. - 10, the surface agents use is generally available, not beta or preview. Local packages, SDKs, frameworks and stdio MCP servers. - 20, installs from an official package with supported runtimes stated. - 25, a public CI and test suite, passing on the default branch. - 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered). - 15, semver discipline and breaking changes called out in a changelog. - 15, version 1.0 or later, or declared stable. Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors. ## 2. Payments & pricing, 30 out of 100, up to 8.8 more on the total Why it scored 30: Payment platforms take the highest step that applies on the 40-point protocol line. No x402, MPP or L402 was found on Paid's own API or for its merchants, whose customers pay by card or bank transfer through a connected Stripe account (0). Plan prices are public at $0, $300, $600 and $1,000 a month by billings tier, with no per-unit price (10). The Free plan covers up to $100,000 of billings a year and the page says no credit card is required (20). A person signs up at app.paid.ai and creates the first key in the dashboard (0). The checklist (https://www.anchorterminal.com/benchmark/#checklist-payments): The published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/). - 40, a machine payment protocol (x402, MPP or L402) on the tool's own endpoints. 10 to 30 when it covers only some endpoints or only goes through a third party, and the note says which. - 20, per-call or per-unit pricing published without a login. 10 for public plan-only pricing, 0 for "contact sales" or prices behind a login. - 20, a free tier or trial that doesn't need a card. - 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API). Payment platforms and agent wallets rarely charge for their own API over a machine protocol, so the first line has steps for them, and the highest one that applies counts. 40 when x402, MPP or L402 runs on all their own endpoints, 30 when it runs on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, and 0 when the only protocol is their own. Merchant acceptance sits above a facilitator because the platform's own customers can charge agents through it, while a facilitator settles for sellers who wire up the protocol themselves. The counter-argument (a facilitator does more for the protocol as a whole) has a point. Each note says which step applied. Open-source software you run yourself is scored on its hosted or paid option if it has one. A free, self-hosted package with nothing to buy gets 20, 20 and 20 for the last three lines, and 0 to 40 for the first only if it ships a payment protocol. ## 3. Security & auth, 65 out of 100, up to 6.1 more on the total Why it scored 65: The MCP server uses OAuth 2.1 with PKCE and client registration, and tools run with the user's role. API keys are full-access or restricted by scope, and customer keys are bound to one customer and one route, expire on request, can be revoked and are stored as a hash. No full scope list was found (27 of 30). Viewers are read-only through MCP and restricted keys limit a workload. No confirmation step for deletes was found (14 of 20). The API returns customer-supplied names and metadata, and the MCP page warns about answers the tools did not produce, with no prompt-injection guidance (7 of 15). An audit log names the user or the key's creator, and keys show `lastUsedAt` (11 of 15). security.txt returned 404, the DPA gives a reporting address and lists TLS 1.2 and AES-256, and no certification or bounty was found in the pages read. trust.paid.ai was not read (6 of 20). The checklist (https://www.anchorterminal.com/benchmark/#checklist-security): - 0 to 30, the credential model. 30 for OAuth 2.1 with scopes, or scoped and revocable keys with rotation. 20 for plain revocable API keys. 10 for one all-powerful key. 10 off when a secret can travel in a URL query string as a documented option. - 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions. - 0 to 15, prompt-injection posture where the tool returns untrusted content (documented mitigations or guidance). A tool that returns no untrusted content gets 10. - 0 to 15, audit logs or per-call visibility for the operator. - 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public. Models are read for retention, whether API data trains models (and whether that's off by default), zero-retention options and certifications. Frameworks for telemetry defaults, approval hooks, guardrails and sandboxing. ## 4. Schema & documentation, 64 out of 100, up to 5.9 more on the total Why it scored 64: The docs index says an OpenAPI 3.1 file is at /v-2/openapi.json and /v-2/openapi.yaml, and both returned 404 on 9 October 2026. Each reference page has a typed Markdown twin and the SDKs are generated from the specification, so 10 of 25. llms.txt per version and Markdown for every page (10). The pages read state purpose and limits well on signals and customer keys and tersely on list calls (15 of 20). Inputs carry enums, defaults and required flags, with free-form maps for `data` and `metadata` (12 of 15). Examples use placeholder values, and errors share one `error`, `code`, `details` shape with named codes in the guides (10 of 15). Docs and paths are versioned v1 and v2. No API changelog was found (7 of 15). The checklist (https://www.anchorterminal.com/benchmark/#checklist-schema): APIs and MCP servers. - 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool). - 10, llms.txt or Markdown docs served for agents. - 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference. - 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs. - 0 to 15, examples and documented error responses. - 15, versioning and a public changelog. Models are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference. ## 5. Agent ergonomics, 70 out of 100, up to 4.9 more on the total Why it scored 70: Graded on the REST API. List calls take `limit` and `offset` and return `total` and `hasMore`, with no field selection, and the MCP server's tool count could not be read (12 of 25). Customers filter by name, status, creation state, dates and external ID (17 of 20). Errors return a stable `code`, and the two kinds of 429 are told apart by it (15 of 20). Signals take an `idempotencyKey`, many resources have upsert or external-ID routes and order amendments have a preview call. MCP annotations were not checked (12 of 20). Official SDKs in five languages and a CLI, with short required lists on the calls read (14 of 15). The checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics): - 0 to 25, context cost. For MCP, the number and size of the tool definitions (25 for ten or fewer compact tools, 15 for 11 to 30, 5 for more than 30, plus up to 10 back for toolsets, dynamic loading or read-only subsets). For APIs, whether responses can be sized (field selection, limits, summaries). - 20, pagination, filtering and output-size controls. - 20, actionable, documented error responses, codes and messages an agent can recover from. - 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations. - 15, sensible defaults, few required parameters, and official SDKs in at least two languages. Models are read for tool use, structured output, prompt caching, context length, batch and SDKs. Frameworks for how much code and how many defaults a tool-calling agent with MCP needs. ## 6. Transparency & trust, 51 out of 100, up to 4.3 more on the total Made of editorial 55, provenance 47. Why it scored 51: The platform is closed under a published Master Subscription Agreement, and the Node and Python SDKs are MIT. The Go SDK repository has no licence file (17 of 30). The agreement, privacy policy and DPA agree with each other on a 30-day retrieval window, deletion from backups within 90 days, hosting on AWS US-East-1 and no training of general-purpose models on customer data. The agreement is undated (22 of 30). Six operations are marked deprecated with a named replacement and no removal date, and no deprecation policy was found (6 of 20). The DPA names AWS and its region and points to trust.paid.ai for the full sub-processor list, which was not read (10 of 20). The checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency): - 0 to 30, source availability and licence clarity. 30 for open source under an OSI licence, 15 for closed with clear terms, 0 for unclear terms. - 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors). - 0 to 20, a deprecation policy or notices with dates. - 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted). The other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two. Provenance checks not met in full (half of this category, computed from checked facts): - Domain age: paid.ai, registered 2017-12-16 (8 years) (11 of 15) - Endpoint on the vendor's domain: api.agentpaid.io is not on paid.ai (0 of 15) - Terms of service: read, states 5 of the 7 things a reader expects, and has 1 clause that costs points (6.3 of 10) - Privacy policy: read, states 7 of the 8 things a reader expects (9.3 of 10) - Status page: not found (0 of 10) - Changelog: not found (0 of 10) - security.txt: not found (0 of 10) ## 7. Maintenance & community, 77 out of 100, up to 2 more on the total Why it scored 77: The CLI `@paid-ai/cli` 1.2.0 was published on 15 September 2026, 24 days before the check (30). Five releases in 90 days, CLI 1.1.0 and 1.2.0, Node SDK 1.8.0, Python SDK 1.12.0 and Go SDK v1.1.0 (20). No public API changelog was found. Support is by email, with Slack on paid plans, and the issue queues were not read (5 of 15). Current official SDKs in five languages (15). The SDK repositories carry CI workflows and lockfiles. CI results were not read, and the Go module still declares go 1.13 (7 of 10). The checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance): - 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older. - 20, at least three releases or dated changelog entries in the last 90 days. - 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15. - 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models). - 10, package health, current dependencies and CI. Models are read for deprecation notice periods and model churn rather than release counts. ## What we couldn't check What we couldn't read counted as absent. Publishing it on a page a plain HTTP fetch can read (not only in a browser) lets the next check count it. - unchecked: trust.paid.ai (the sub-processor list and any certifications). The website terms forbid automated access at scale to Paid's sites, so we stopped at the pages already fetched and the docs' published agent files - unchecked: the paid.ai home page and product pages, for the same reason. A status page link may sit on a page we did not read. None is in the pricing page footer or the docs index - unchecked: the OpenAPI document. Both addresses in the docs index returned 404, so operation counts come from the reference index and schema quality from six reference pages - unchecked: the MCP server's tool names, count, schemas and annotations. The server needs a sign-in and its source is not public - unchecked: who publishes the `paid` package on PyPI that the quickstart tells users to install. The PyPI project page answered with a client challenge. The SDK's README and `pyproject.toml` name `paid-python` - unchecked: GitHub stars, issue queues and CI results for the SDK repositories, and PyPI download counts - unchecked: the registration date of agentpaid.io. RDAP returned no events - unchecked: the list of API key scopes. Three scope names appear in the pages read and no page listing them all was found - The lead named the Python package `paid`. The SDK repository names it `paid-python`. The lead also listed only a docs MCP server, and Paid runs a product MCP server at mcp.agentpaid.io, a CLI, and Ruby and Java SDKs - The plan prices were decoded from the transform values of the pricing page's digit columns, not read as plain text ## Weaknesses - The docs index advertises an OpenAPI 3.1 file at two addresses, and both returned 404 on 9 October 2026 - No status page, uptime figure or SLA percentage was found. The agreement promises commercially reasonable efforts only - Rate limits are described as a token bucket per organisation with no published bucket size or refill rate - No API changelog or deprecation policy was found. Six operations are marked deprecated with no removal date - The API and MCP server run on agentpaid.io, a second domain, and collecting card or bank payments needs the builder's own connected Stripe account ## What costs an agent a turn today The notes we give agents before they call it. Each one is a workaround an agent shouldn't need. - Call `https://api.agentpaid.io/api/v2` with `Authorization: Bearer `. A person creates the first key at app.paid.ai. No programmatic signup was found - Give every signal a stable `idempotencyKey` and resend the same key on retry. Signals without one are not deduplicated - Send at most 500 signals a request. A 429 rejects the whole batch, so wait for `Retry-After` and resend it all - Read the `code` field on a 429. `CONCURRENCY_LIMIT_EXCEEDED` carries no `Retry-After` and clears when an analytics query finishes - Install the Python SDK as `paid-python`, the name in its README and `pyproject.toml`. The quickstart's `pip install paid` names a package we could not check ## When it's done Send what changed and where it's published as a dispute (https://www.anchorterminal.com/builders/#disputes, or `POST https://www.anchorterminal.com/api/v1/contact` with `"kind": "dispute"`). Disputes are answered in public, and the listing is checked again by the same checklist. Paying for an audit or a listing claim changes nothing here.