Prefactor

# Fix list: Prefactor

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/prefactor, the October 2026 research run, assessed 8 October 2026. Grade B, 65.6 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 Prefactor: 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, 56 out of 100, up to 7.7 more on the total

Why it scored 56: Bearer tokens scoped to the account or one agent deployment, with a `read_only` or `full_access` role, an expiry, and suspend, reactivate and revoke. The role appears only in the OpenAPI document, the default is full access and the default expiry is two years (27 of 30). Read-only and deployment tokens give least privilege. Deletes, including `account/{id}/implode`, have no confirmation step in the API, and the docs say all account members share one access level (12 of 20). Span payloads hold whatever the instrumented agent saw, the API returns marked sensitive values unless `redacted: true` is passed, and we found no prompt-injection guidance for agents reading them (4 of 15). Tokens show last use and creator, and quality and alert changes are written as spans. No log of API calls for the operator was found (7 of 15). A disclosure address with a 24-hour acknowledgement promise. No security.txt, no bounty, no advisories page, and SOC 2 Type II is described as in progress (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.

## 2. Payments & pricing, 40 out of 100, up to 7.5 more on the total

Why it scored 40: No machine payment protocol (0). Per-span overage prices and plan prices are published without a login, $0.0025 a span and $49 or $199 a month (20). The Dev plan is free with 5,000 spans a month and the page says no credit card is required (20). A person signs in through the browser to create the first token. `POST /api_token` mints more only with an existing token (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. Transparency & trust, 42 out of 100, up to 5.1 more on the total

Made of editorial 34, provenance 50.

Why it scored 42: A closed service with MIT SDKs. The only published terms cover the website, so the terms for the product are unclear (10 of 30). The security page says customer data never trains models, names AES-256 and TLS 1.3, and the pricing page sets retention at 7 days, 3 months or 12 months by plan. The privacy policy is a generated template whose regional sections place storage in the United Kingdom and in Canada, which does not agree with the AWS regions on the security page, and we found no DPA (12 of 30). No deprecation policy. The SDK changelogs mark breaking changes with major versions (4 of 20). Three AWS regions are named as residency options. No sub-processor list was found (8 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: prefactor.tech, registered 2024-10-16 (1 year) (3 of 15)
- Endpoint on the vendor's domain: app.prefactorai.com is not on prefactor.tech (0 of 15)
- Terms of service: not found (0 of 10)
- Privacy policy: published, but our reader couldn't read it (7 of 10)
- security.txt: not found (0 of 10)

## 4. Reliability, 78 out of 100, up to 4.4 more on the total

Why it scored 78: Graded as a hosted service. A public status page at status.prefactor.ai (UptimeRobot) with three monitors and 90 days of daily bars, though we found no link to it from the site or docs (20). All three monitors, the API host among them, show 100 per cent for the 90 days to 8 October 2026 and no status updates. These are plain HTTP checks, not per-endpoint probes (30). Rate limits are described by scope and window with no numbers (3 of 15). 429 carries `Retry-After` and `retry_after_ms`, both SDKs retry with backoff and jitter, and every operation takes an idempotency key (15). The pricing page lists an SLA on Enterprise with no text published (0). The API is version 1.0 and `@prefactor/core` is 1.2.0, with no beta label (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.

## 5. Schema & documentation, 80 out of 100, up to 3.3 more on the total

Why it scored 80: A public OpenAPI 3.0.0 document with 157 operations and an OpenRPC document with 159 WebSocket methods. Body parameters are declared `in: body`, which OpenAPI 3 does not define, so some generators will need adjusting (23 of 25). llms.txt with abridged and full versions of the docs (10). Operations carry one-line summaries and only 4 of 157 have a description. The concept pages are clearer and say what the product does not do (10 of 20). Named ID schemas, enums and bounds such as `page_size` 1 to 100, but span payloads are free-form by design and `start_time` and `end_time` have no schema (11 of 15). A table of 12 error codes, an error union in the spec, curl examples and a runnable WebSocket script (13 of 15). The path is versioned `/api/v1`, the SDKs keep semver changelogs, and the site changelog has 35 dated entries, the latest on 27 July 2026 (13 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.

## 6. Agent ergonomics, 88 out of 100, up to 2 more on the total

Why it scored 88: List calls take `pagination[page_size]` up to 100, summaries and counts are opt-in flags, and one endpoint exports a whole run's context (16 of 25). Offset pagination, sorting and filters by agent, instance, environment, time, purpose and status, with no cursor (18 of 20). Errors carry a stable `code`, a message naming the exceeded scope on 429 and a per-field map on validation failures (20). Every operation accepts `idempotency_key`, and a retried bulk request fails only the reused items. There is no MCP server to annotate (20). Official TypeScript and Python SDKs and a CLI. The Python SDK does not yet support sensitive encoding (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.

## 7. Maintenance & community, 82 out of 100, up to 1.6 more on the total

Why it scored 82: `@prefactor/core` 1.2.0 was published on 13 September 2026, 25 days before this check (30). Four releases of it since 10 July (0.5.1, 1.0.0, 1.1.0, 1.2.0), plus five changelog entries in late July (20). A closed service with a public changelog whose latest entry is 27 July 2026, a Slack community and public SDK repositories with 4 open issues and pull requests on the TypeScript one. We did not read reply times (9 of 15 available to a closed service). Current official SDKs in two languages (15). CI passed on the TypeScript main branch on 13 September and weekly scheduled runs passed to 8 October. The npm release workflow failed on 13 September, and the Python SDK's last release is 13 August (8 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.

## Deductions

Each comes off the total. A fixed and documented problem counts for less at the next check.

- 2026-10-08. The pricing page lists 'Hold, approve or block in real time' and PII detection on every plan, and the security page lists manual approval workflows and fine-grained RBAC. The vendor's own docs say crossing a risk threshold does not block or terminate a run, that Prefactor does not detect sensitive data, and that accounts have no role-based permissions. The OpenAPI and OpenRPC documents have no approval operation. Counted as a misleading claim (-3). The docs may lag the product, since the API has hosted judge operations they do not describe (https://prefactor.tech/pricing, https://prefactor.tech/security, https://docs.prefactor.ai/llms-full.txt).

## 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.

- Whether the hold, approve and block controls on the pricing page exist in the product. The docs deny automatic blocking and the API documents have no approval operation.
- Whether the hosted judge operations in the OpenAPI document are available to every account. The docs do not describe them.
- unchecked: the main body of the privacy policy. The page draws it by script and we could render only its regional sections from the feed.
- unchecked: whether signup shows a service agreement. We did not open the signup flow.
- unchecked: reply times on GitHub issues and in the Slack community.
- The published rate-limit numbers, the SLA text and any sub-processor list or DPA were not found.
- The OpenTelemetry ingest named in the lead and in one changelog entry has no documented endpoint.

## Weaknesses

- The pricing page lists hold, approve or block and PII detection on every plan, while the docs say a risk threshold blocks nothing and that Prefactor does not detect sensitive data
- No approval operation exists in the OpenAPI or OpenRPC documents. The documented control is a cooperative terminate signal the agent's own code must obey
- The published terms of service cover the website. No service agreement, DPA, sub-processor list or SLA text was found
- Rate limits are described by scope with no numbers
- No security.txt and no certification held. The security page says SOC 2 Type II is in progress

## 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 requests to `https://app.prefactorai.com/api/v1` with `Authorization: Bearer <token>`. The API host is not on the `prefactor.tech` or `prefactor.ai` domains
- Ask for a deployment-scoped token, or an account token created with `role` set to `read_only`, since the default role is `full_access` with a two-year expiry
- Pass `redacted: true` on span list and detail queries, because the API returns marked sensitive values unless asked otherwise
- On 429 wait for `retry_after_ms`. Limits use one-minute windows and the SDKs' default retries ignore the server's hint
- Treat terminate as a request. Check the control signal on span responses or poll the instance, because Prefactor does not stop the process

## 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.