Greenhouse

# Fix list: Greenhouse

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/greenhouse, the October 2026 research run, assessed 7 October 2026. Grade B, 64.8 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 Greenhouse: 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. Payments & pricing, 0 out of 100, up to 12.5 more on the total

Why it scored 0: Read with the hosted rubric. No x402, MPP or L402 (0). No public prices. The pricing page names Core, Plus and Pro and says pricing is customised (0). No free tier or trial found, and the sandbox is a Pro-tier feature (0). Credentials are created by a person inside a customer account, and the MCP server's dynamic client registration still needs a paying organisation and a user's consent in a browser (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.

## 2. Agent ergonomics, 63 out of 100, up to 6 more on the total

Why it scored 63: `per_page` runs from 1 to 500 and v3 no longer embeds child records in parents, but no field selection was found (15). Cursor pagination through the `Link` header, with filters for ids, parent ids, status, `created_at` and `updated_at` (20). Errors return a `message` and an `errors` array that names the fault, such as a cursor combined with other parameters, though reference pages don't show them (13). No idempotency keys found. Moving an application needs the current stage as a guard, and the MCP server blocks DELETE and confirms five actions. Its tool annotations couldn't be read without an account (8). Creating a candidate needs two fields, but no official SDKs were found (7).

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.

## 3. Reliability, 71 out of 100, up to 5.8 more on the total

Why it scored 71: Read with the hosted rubric, for the Harvest v3 API. Statuspage at status.greenhouse.io with a Greenhouse Harvest API component and per-silo components (20). From 9 July to 7 October 2026 the page lists 14 incidents, three marked major. One was a partial disruption on Silo 1 on 5 August, opened at 02:18 UTC with a fix at 03:56 and closed at 13:29, covering scorecards, notes, emails and offer letters. The others were 26 minutes on Silo 201 on 2 September and 38 minutes of MyGreenhouse job search on 21 September. None names the Harvest API, so we scored between minor-only and one major (15). The v3 limit is a fixed 30-second window whose size arrives in `X-RateLimit-Limit`, with no number published beyond an example of 75 (8). 429 carries `Retry-After` and the docs give backoff guidance, but no idempotency keys were found for writes (10). The SLA commits to 99.5 per cent for Greenhouse Recruiting with service credits and does not name the API (8). Harvest v3 is generally available, while the MCP server is in open beta (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.

## 4. Schema & documentation, 79 out of 100, up to 3.4 more on the total

Why it scored 79: Every v3 reference page has a Markdown twin with an OpenAPI 3.1 definition for that endpoint. No single downloadable spec was found (20). llms.txt indexes 179 reference pages and the guides, all served as Markdown (10). Descriptions state purpose, required fields and limits, for example that stage moves are not handled by the update call and that `from_stage_id` guards against stale moves (16). Request bodies set `additionalProperties` to false with required fields, enums and length limits (13). Success examples are given, but reference pages list 401, 403 and 422 without bodies, and error bodies appear only in the guides (8). The path is versioned and the changelog had ten entries in 27 days, titled by commit hash with relative dates (12).

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. Maintenance & community, 65 out of 100, up to 3.1 more on the total

Why it scored 65: The Harvest v3 changelog showed an entry about four hours old when read on 7 October 2026 (30). Ten entries in the previous 27 days (20). Closed service with a changelog, monthly release notes that list API changes (the September 2026 notes add the `interview_media` endpoint) and a partner support address. Changelog entries are titled by commit hash, which makes them hard to follow (10). No official SDKs found, and the MCP server is not in the official MCP registry, where a search returned only a third-party scraper (0). No package to assess. The docs repository for the older APIs had commits up to 10 September 2026 (5).

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.

## 6. Security & auth, 85 out of 100, up to 2.6 more on the total

Why it scored 85: OAuth 2.0 with a scope per endpoint, client credentials or the authorisation code grant, one-hour partner access tokens and secret rotation that keeps the old secret for up to a week. The MCP server adds PKCE and dynamic client registration (30). Read-only access is possible by scope, the MCP defaults to seven read-only scopes, blocks every DELETE and asks for confirmation on five actions. All list endpoints need a Site Admin as the authorising user (16). The MCP security FAQ covers prompt injection (structured tool calls only, permission-bound and rate-limited), with no guidance on treating candidate text in responses as untrusted (9). The Audit Log API records Harvest and MCP calls with user, client and arguments for thirty days, as a paid add-on (12). SOC 2 Type II, ISO 27001, 27701 and 42001, a HackerOne bug bounty and annual penetration tests. security.txt expired on 1 February 2026 (18).

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.

## 7. Transparency & trust, 79 out of 100, up to 1.8 more on the total

Made of editorial 67, provenance 90.

Why it scored 79: Closed service with a public Master Subscription Agreement last updated 1 February 2026, and an Apache-2.0 docs repository (15). The agreement queues customer data for deletion 90 days after termination, a DPA is published, and the MCP FAQ says the server stores no customer data. The privacy policy of 28 May 2026 says personal information may be used to develop and train AI models and gives no retention periods (20). The removal of Harvest v1 and v2 was announced for 31 August 2026 with migration guides, but five weeks later the banner still read as future and v1 still answered. No written deprecation policy with a notice period was found (12). A public sub-processor list with purposes and locations, AWS hosting in the United States, Germany and Australia, and the MCP server stated as US-only (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)
- Privacy policy: read, states 8 of the 8 things a reader expects, and has 1 clause that costs points (8 of 10)
- security.txt: published but past its Expires date (5 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: the numeric Harvest v3 rate limit. The guide gives a 30-second window and an example header of 75 but no fixed figure
- unchecked: the MCP server's tool list, count, input schemas and annotations, which need an authorised customer account
- unchecked: whether Harvest v1 and v2 still serve data after the announced removal date of 31 August 2026. An unauthenticated request to /v1/candidates returned 401 on 7 October 2026
- unchecked: most Harvest v3 changelog entry texts and the migration guides, because harvestdocs.greenhouse.io answered several requests with a bot check
- unchecked: the Data Processing Addendum text and the AI terms the Master Subscription Agreement refers to
- unchecked: GitHub stars for grnhse/greenhouse-api-docs and the registration date of greenhouse.io
- Not found in the reviewed documentation: idempotency keys, field selection, official SDKs, a single OpenAPI file, and a deprecation policy with a notice period

## Weaknesses

- No public price, free tier or trial. Pricing is customised through sales, and the sandbox is a Pro-tier feature
- The v3 rate limit is a 30-second window with no published number. The limit arrives in `X-RateLimit-Limit`
- No idempotency keys found, and reference pages list 401, 403 and 422 without error body examples
- All list endpoints need a Site Admin as the authorising user, per the authentication guide
- No official SDKs found, and the MCP server is in open beta and absent from the official MCP registry

## 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.

- Request a token from https://auth.greenhouse.io/token with client credentials, then send it as a Bearer token to https://harvest.greenhouse.io/v3. Tokens expire, so repeat on 401
- Put filters and `per_page` on the first list request only. A `cursor` must be the only query parameter, or the call returns 422
- Send the current stage as `from_stage_id` when moving an application. It guards against stale moves. Other writes have no idempotency key, so check before retrying
- Read `X-RateLimit-Remaining` on every response and wait for `Retry-After` on 429. The window is 30 seconds
- Expect empty list results or 403 unless the token's user is a Site Admin with the needed permissions

## 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.