TaxJar

# Fix list: TaxJar

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/taxjar, the October 2026 research run, assessed 8 October 2026. Grade C, 57.6 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 TaxJar: 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. Security & auth, 43 out of 100, up to 10 more on the total

Why it scored 43: One API token per use, generated in the web app and sent in the `Authorization` header, never in the URL. It has no scopes and covers every endpoint. The docs mention generating a new token, but we found no statement on how many tokens an account can hold or how one is revoked, so we scored between a single all-powerful key and plain revocable keys (15 of 30). No read-only token and no confirmation step on deleting transactions or customers. The sandbox takes a separate token, and user roles exist for web app logins on Professional (4 of 20). Responses carry tax figures, jurisdiction names and the account's own stored customer and order data, with little third-party text (9 of 15). Plans & Billing shows transaction counts, and no per-request API log was found in the reviewed documentation (3 of 15). SOC 2 Type 2 per the security page, with yearly re-audits announced in March 2021, annual penetration testing and AES-256 at rest. No security.txt on TaxJar hosts and no disclosure policy or bounty page for TaxJar was found (12 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.

## 2. Payments & pricing, 30 out of 100, up to 8.8 more on the total

Why it scored 30: Read with the hosted rubric. No x402, MPP or L402 (0). Plan prices are public, Professional from $99 a month at 200 orders, with a stated conversion of ten calculations to one order. Higher tiers sit on a selector, and the Premium plan named in the rate-limit docs has no public price. That is plan pricing (10 of 20). A 30-day trial with Professional plan access and the API needs no card (20). A person has to sign up in a browser and generate the token in the web app (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. Schema & documentation, 48 out of 100, up to 8.5 more on the total

Why it scored 48: No OpenAPI spec was found. The taxjar/openapi repository on GitHub holds only a README, last changed in May 2020. The Node client ships TypeScript parameter and return types, which we counted as a partial contract (4 of 25). No llms.txt on developers.taxjar.com or www.taxjar.com, both 404 (0). The reference is one page covering 21 operations, with notes on sourcing, nexus and exemptions, and guides on product tax codes, marketplaces and avoiding unnecessary calls. Few entries say when not to use an endpoint (12 of 20). Parameter tables give a type and required, conditional or optional for each field, with allowed values for `exemption_type` in prose. Constraints such as amount arithmetic are enforced only on newer API versions (9 of 15). Request and response examples for every operation in cURL and seven client languages, an error code table, causes and fixes for 400, 401, 403, 406 and 422, and a sandbox header that mocks errors (13 of 15). Versions are dated and set by the `x-api-version` header, with a public changelog whose newest entry is 24 January 2022 (10 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.

## 4. Agent ergonomics, 61 out of 100, up to 6.3 more on the total

Why it scored 61: Responses are small JSON objects and list calls return IDs only, but there is no field selection and `summary_rates` returns every region at once (12 of 25). Order and refund lists filter by `transaction_date`, a date range and `provider`. No pagination or page size is documented, and the customer list takes no parameters (8 of 20). Errors are JSON with `error`, `detail` and `status`, which we confirmed on a 401, and five codes have documented causes and fixes (15 of 20). Calculations have no side effects beyond metering. Orders, refunds and customers use caller-chosen IDs, so a repeated create returns 422 and can be turned into a PUT. No idempotency header exists (13 of 20). A calculation needs `to_country`, `to_zip`, `to_state` and `shipping`, with nexus read from the account, and official clients cover seven languages, though the Ruby gem dates from August 2022 and the Java and Go repositories were last pushed in 2023 (13 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.

## 5. Reliability, 84 out of 100, up to 3.2 more on the total

Why it scored 84: Read with the hosted lines and scored on the public REST API at api.taxjar.com, the surface an agent would call. TaxJar has no MCP server. status.taxjar.com on Statuspage lists seven components with incident history (20). No incident in the 90 days to 8 October 2026. The last two were an integration fault on 26 March 2026 and a degraded API on 28 February 2026, when calls with a `customer_id` returned 5xx for about 4 hours 50 minutes (30). Rate limits with numbers, 10,000 requests a minute on Professional and 25,000 on Premium (15). The docs list 429 and advise caching, calculating only in nexus states and falling back to summarised rates. No Retry-After header or backoff guidance was found. Writes are keyed on a caller-chosen `transaction_id`, a duplicate create returns 422 and the docs direct a PUT (9 of 15). No uptime SLA was found. The published SLA covers support response times only, and the terms supply the API as is (0). The API is at v2 and generally available (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.

## 6. Maintenance & community, 64 out of 100, up to 3.2 more on the total

Why it scored 64: A closed service, so we read releases from the official clients. The Node client 5.0.0 reached npm on 2 September 2026, 36 days before the check (20 of 30). In the 90 days to 8 October 2026 there were also .NET 5.0.0 on NuGet on 4 August, the Magento extension 4.0.2 on 3 August and WooCommerce plugin tags on 2 September, 9 September and 5 October. None is an API change, and the API changelog has no entry since 24 January 2022 (20). Support is by email with a published response time of one business day on Professional, though the API licence agreement promises no API support. GitHub issues are open on the client repositories, 8 on Node and 19 on .NET, and we didn't measure reply times (8 of 15). Seven official clients, four released since April 2025. The Ruby gem dates from August 2022, and the Java and Go repositories were last pushed in 2023 (11 of 15). The Node repository runs CodeQL only, with no test workflow in public CI, and its 5.0.0 release dropped a deprecated dependency (5 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.

## 7. Transparency & trust, 71 out of 100, up to 2.5 more on the total

Made of editorial 55, provenance 87.

Why it scored 71: The service is closed under terms of service and an API licence agreement last updated 14 May 2026. The clients are MIT (15 of 30). The privacy policy dates from 7 January 2022 and gives no fixed retention period. The data processing agreement of 14 May 2026 promises deletion or return of personal data after termination at the customer's choice, and notice of a data incident without undue delay. The documents agree but name no periods (18 of 30). No deprecation policy was found. Three dated API versions remain accepted, and the terms promise 30 days' notice of changes to terms and prices. International calculation was closed to new customers with no dated notice in the reviewed pages (5 of 20). The sub-processor list, updated 26 June 2026, names seven processors with purpose and country, and the privacy policy says the service is hosted in the United States (17 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):

- Terms of service: read, states 6 of the 7 things a reader expects, and has 1 clause that costs points (7.1 of 10)
- security.txt: not found (0 of 10)

## 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: how many API tokens an account can hold and how a token is revoked or rotated. The docs say only that a new token can be generated at Account > TaxJar API, and we didn't sign up.
- unchecked: prices for Professional tiers above 1,000 orders and for the Premium plan. The pricing page shows the 200-order tier and a selector, and Premium appears only in the rate-limit docs.
- unchecked: whether 429 responses carry a Retry-After header. The docs don't mention one and we didn't trigger a 429.
- unchecked: whether the web app keeps a per-request API log. The docs mention only transaction counts under Plans & Billing.
- unchecked: whether Stripe's HackerOne programme covers taxjar.com hosts. No disclosure policy was found on TaxJar's own pages.
- unchecked: current versions of the Java and Go clients on their registries. Their repositories were last pushed in May and October 2023.
- unchecked: the role names and permissions under Users & Roles. The help centre article shows them only in screenshots.
- The product page claims 99.999 per cent historical uptime and sub-20ms response times. These are vendor claims and no uptime SLA was found.
- The sandbox article, last updated 28 September 2021, says address validation is unsupported there. We didn't test the sandbox.
- No vendor MCP server was found, and the official MCP registry has no taxjar entry.
- The Node client's changelog dates 5.0.0 to 3 February 2026, while npm shows it published on 2 September 2026. We used the npm date for the last release.
- The API has no filing endpoint, so `tax.filing` was left out of the capabilities although TaxJar sells AutoFile in its web app.

## Weaknesses

- No OpenAPI spec or llms.txt was found. The taxjar/openapi repository on GitHub holds only a README
- The API changelog's last entry is 24 January 2022, and no deprecation policy was found
- Calculation covers the United States only for new accounts. The docs point other countries to Stripe Tax
- API tokens carry no scopes or read-only mode, and the same token can delete transactions and customers
- Calls that return an error still count as a tenth of an order against the plan, per the help centre

## 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.

- Send `x-api-version: 2022-01-24` on every call. Accounts opened before 3 February 2022 default to an older version with weaker validation
- Pass `nexus_addresses[]` or keep nexus states in the account. Without nexus the API returns zero tax and reports no error
- Create orders with your own `transaction_id`. On 422 the order already exists, so send a PUT to update it
- Post JSON with `Content-Type: application/json`. Form-encoded arrays such as `line_items[]` can return 406
- Cache calculations and skip states without nexus. Each calculation or rate lookup counts as a tenth of an order, errors included

## 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.