{
  "fixes": {
    "slug": "coreweave",
    "name": "CoreWeave",
    "listing": "https://www.anchorterminal.com/tools/coreweave",
    "markdown": "# Fix list: CoreWeave\n\nFrom Anchor Terminal's listing at https://www.anchorterminal.com/tools/coreweave, the October 2026 research run, assessed 8 October 2026. Grade C, 61.5 out of 100.\n\nThis 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.\n\nFor a coding agent working on CoreWeave: 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.\n\n## 1. Reliability, 50 out of 100, up to 10 more on the total\n\nWhy it scored 50: Graded as a hosted service on the CKS and platform APIs. Status.io page at status.coreweave.com with components per data centre, an RSS feed and email, webhook and Slack subscriptions (20). The history page is drawn by script and the feed holds ten items back to 25 September 2026, so only the last two weeks were readable. In that window the Cloud Console returned 404 for about 40 minutes on 6 October, US-CENTRAL-09A lost network connectivity from 06:22 to 08:37 UTC on 5 October, and object storage in the same zone ran at reduced IOPS for about two hours on 26 September. One major regional outage (10). No request rate limits found for api.coreweave.com, only gateway size limits of about 50 KB and 100 MB, and quotas are set per customer through support (0). No 429 or backoff guidance and no idempotency keys found. Kubernetes resources are declarative, so re-applying a manifest is safe (5 of 15). The terms set an objective of 99.9 per cent across multiple regions and 99 per cent in one, with financial credits (10). CKS is sold as a general service but its cluster API is v1beta1 and the Node Pool resource v1alpha1 (5 of 10).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):\n\nHosted APIs, MCP servers, models and platforms.\n\n- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).\n- 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.\n- 15, rate limits documented with numbers.\n- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.\n- 10, an SLA published for any paid tier.\n- 10, the surface agents use is generally available, not beta or preview.\n\nLocal packages, SDKs, frameworks and stdio MCP servers.\n\n- 20, installs from an official package with supported runtimes stated.\n- 25, a public CI and test suite, passing on the default branch.\n- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).\n- 15, semver discipline and breaking changes called out in a changelog.\n- 15, version 1.0 or later, or declared stable.\n\nProtocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors.\n\n## 2. Payments \u0026 pricing, 20 out of 100, up to 10 more on the total\n\nWhy it scored 20: No machine payment protocol (0). Per instance-hour prices for GPUs, CPUs, storage and networking are published without a login, with some models quoted by sales (20). No free tier or self-serve trial found. Trials run through the sales-led ARENA programme (0). An organisation has to be approved by the sales team and a person creates the token in the console (0).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-payments):\n\nThe published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/).\n\n- 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.\n- 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.\n- 20, a free tier or trial that doesn't need a card.\n- 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API).\n\nPayment 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.\n\nOpen-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.\n\n## 3. Agent ergonomics, 58 out of 100, up to 6.8 more on the total\n\nWhy it scored 58: Graded on the platform API, the Kubernetes API and the Mission Control MCP server together. The MCP server documents 38 tools, most of them small read-only queries with limits, and every tool also takes an optional `user_prompt` (12 of 25). The Dedicated Inference API pages with `maxPageSize` and `pageToken`. The cluster list returns everything, with `returnPartialSuccess` for unreachable zones (14). Errors follow the google.rpc shape with a numeric code and a message, and support articles explain common failures (12). No idempotency keys, no MCP read-only or destructive annotations documented, and `coreweave_kubectl_apply` ignores `dry_run` and `plan_token` and applies the manifest (8). Generated clients on Buf for Go, Python, TypeScript, Java and others, plus a Terraform provider. Creating a cluster needs a zone, version and VPC (12).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics):\n\n- 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).\n- 20, pagination, filtering and output-size controls.\n- 20, actionable, documented error responses, codes and messages an agent can recover from.\n- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.\n- 15, sensible defaults, few required parameters, and official SDKs in at least two languages.\n\nModels 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.\n\n## 4. Security \u0026 auth, 74 out of 100, up to 4.6 more on the total\n\nWhy it scored 74: API access tokens are user-scoped, carry an expiry, are shown once and can be revoked in the console. They inherit the user's IAM roles and have no scopes of their own. OIDC workload identity federation is available for CKS workloads. Tokens travel in the Authorization header only (25). IAM has a Viewer and an Admin role per service, and the MCP write tools are off unless CoreWeave enables them for an organisation. There is no confirmation step, and the apply tool ignores `dry_run` (15). The MCP server returns log lines and documentation text, which are untrusted content, and no injection guidance was found (5). Kubernetes audit logs have a Grafana dashboard, Telemetry Relay forwards audit logs to HTTPS or S3, and support staff access is controlled and logged (15). Vulnerability disclosure policy with safe harbour, trust centre badges for SOC 2 Type II, ISO 27001, 27017 and 27018, and a security advisories page whose latest entry is July 2024. No security.txt and no bug bounty found (14).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-security):\n\n- 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.\n- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.\n- 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.\n- 0 to 15, audit logs or per-call visibility for the operator.\n- 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public.\n\nModels 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.\n\n## 5. Schema \u0026 documentation, 79 out of 100, up to 3.4 more on the total\n\nWhy it scored 79: OpenAPI 3.0 specs for the CKS, VPC, Object Storage, Dedicated Inference, Capacity Finder and Telemetry Relay APIs are embedded in the reference pages, and the same services are published as Protobuf on the Buf Schema Registry. The raw .yaml paths in llms.txt redirected our reader to a docs login (25). llms.txt at docs.coreweave.com and every page served as Markdown (10). Operations and MCP tools state their purpose and inputs, with few lines on when not to use them (12). Required fields are marked and the Node Pool reference lists enums, but the CKS spec has none and instance types and versions are plain strings (9). Each operation page has a curl example. Errors are one generic `code`, `message`, `details` shape with no per-operation codes (8). Versioned paths and a dated changelog of 168 entries since December 2024 (15).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-schema):\n\nAPIs and MCP servers.\n\n- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).\n- 10, llms.txt or Markdown docs served for agents.\n- 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.\n- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.\n- 0 to 15, examples and documented error responses.\n- 15, versioning and a public changelog.\n\nModels 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.\n\n## 6. Transparency \u0026 trust, 77 out of 100, up to 2 more on the total\n\nMade of editorial 71, provenance 83.\n\nWhy it scored 77: Closed service with published terms. They were last modified on 30 June 2022 and say they may be amended without notice (15). The terms commit to deleting customer data after account closure within a 30-day recovery period and at most 180 days, there is a data processing agreement, and the privacy policy of 24 February 2026 gives no fixed retention period for personal data (22). The changelog gives dated deprecations, for example operating system and driver versions from 29 July 2026, and the maintenance policy promises 24 hours' notice for critical work, though it can change without notice (14). Three sub-processors are named with their country and 30 days' notice of additions, and regions and availability zones are documented (20).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency):\n\n- 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.\n- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).\n- 0 to 20, a deprecation policy or notices with dates.\n- 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted).\n\nThe other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two.\n\nProvenance checks not met in full (half of this category, computed from checked facts):\n\n- Domain age: coreweave.com, registered 2019-04-01 (7 years) (11 of 15)\n- Terms of service: read, states 6 of the 7 things a reader expects, and has 1 clause that costs points (7.1 of 10)\n- security.txt: not found (0 of 10)\n\n## 7. Maintenance \u0026 community, 80 out of 100, up to 1.8 more on the total\n\nWhy it scored 80: The changelog's latest entry is 2 October 2026 (30). It has 18 dated entries between 10 July and 2 October 2026 (20). Closed service with a dated changelog, a support centre and status updates by email. No public issue tracker for the platform (10 of 15). Generated clients on the Buf Schema Registry and a Terraform provider tagged v0.24.0 on 11 September 2026, the eighth tag since 14 July. The MCP server isn't in the official MCP registry (12). The provider repository has build, test, acceptance-test, release and Renovate workflows. We didn't check that they pass (8).\n\nThe checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance):\n\n- 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.\n- 20, at least three releases or dated changelog entries in the last 90 days.\n- 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.\n- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).\n- 10, package health, current dependencies and CI.\n\nModels are read for deprecation notice periods and model churn rather than release counts.\n\n## What we couldn't check\n\nWhat 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.\n\n- unchecked: status history before 25 September 2026. The Status.io history page is drawn by script and the feed keeps ten items, so the 90-day record is scored on two weeks.\n- unchecked: the raw OpenAPI .yaml files named in llms.txt redirected to a docs login, so the specs were read as embedded in the reference pages.\n- unchecked: whether the Terraform provider's CI passes on the default branch.\n- unchecked: id.coreweave.com/signup is drawn by script. The docs say the sales team approves each organisation, and we couldn't see whether that page opens an infrastructure account without approval.\n- unchecked: the trust centre's documents sit behind an access request, so the certifications are badges we read, not reports.\n- No request rate limits, billing increment for on-demand nodes or bug bounty were found in the reviewed documentation.\n- The lead lists Serverless Inference as part of this product. It is a Forge service at api.inference.wandb.ai with its own key and plans, so it is noted here and not graded.\n\n## Weaknesses\n\n- No self-serve route. CoreWeave's sales team approves an organisation and emails the activation link before a token can be created\n- No request rate limits, 429 guidance or idempotency keys found for api.coreweave.com in the reviewed documentation\n- The CKS API is versioned v1beta1 and the Node Pool resource and Dedicated Inference API v1alpha1\n- Three incidents between 26 September and 6 October 2026, including a US-CENTRAL-09A network outage of over two hours on 5 October\n- The MCP tool `coreweave_kubectl_apply` ignores its `dry_run` parameter and applies the manifest, per the tool reference\n\n## What costs an agent a turn today\n\nThe notes we give agents before they call it. Each one is a workaround an agent shouldn't need.\n\n- Create an API access token at console.coreweave.com/tokens with an expiry and send it as `Authorization: Bearer` to https://api.coreweave.com\n- Add GPUs by applying a NodePool resource (`compute.coreweave.com/v1alpha1`) to a CKS cluster with `instanceType` and `targetNodes`. `instanceType` can't be changed afterwards\n- Set `targetNodes` to 0 or delete the Node Pool when a job ends. Nodes are whole 8-GPU machines billed by the hour\n- Never pass `dry_run` to the MCP tool `coreweave_kubectl_apply` expecting a preview. The reference says the parameter is ignored and the manifest is applied\n- Use a separate Forge API key and https://api.inference.wandb.ai/v1 for Serverless Inference. A CoreWeave API access token doesn't work there\n\n## When it's done\n\nSend 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.\n",
    "grade": "C",
    "score": 61.5,
    "assessed": "2026-10-08",
    "run": "October 2026 research run",
    "categories": [
      {
        "key": "reliability",
        "name": "Reliability",
        "score": 50,
        "maxGain": 10,
        "reason": "Graded as a hosted service on the CKS and platform APIs. Status.io page at status.coreweave.com with components per data centre, an RSS feed and email, webhook and Slack subscriptions (20). The history page is drawn by script and the feed holds ten items back to 25 September 2026, so only the last two weeks were readable. In that window the Cloud Console returned 404 for about 40 minutes on 6 October, US-CENTRAL-09A lost network connectivity from 06:22 to 08:37 UTC on 5 October, and object storage in the same zone ran at reduced IOPS for about two hours on 26 September. One major regional outage (10). No request rate limits found for api.coreweave.com, only gateway size limits of about 50 KB and 100 MB, and quotas are set per customer through support (0). No 429 or backoff guidance and no idempotency keys found. Kubernetes resources are declarative, so re-applying a manifest is safe (5 of 15). The terms set an objective of 99.9 per cent across multiple regions and 99 per cent in one, with financial credits (10). CKS is sold as a general service but its cluster API is v1beta1 and the Node Pool resource v1alpha1 (5 of 10).",
        "checklist": [
          "Hosted APIs, MCP servers, models and platforms.",
          "- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).\n- 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.\n- 15, rate limits documented with numbers.\n- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.\n- 10, an SLA published for any paid tier.\n- 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.\n- 25, a public CI and test suite, passing on the default branch.\n- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).\n- 15, semver discipline and breaking changes called out in a changelog.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-reliability"
      },
      {
        "key": "payments",
        "name": "Payments \u0026 pricing",
        "score": 20,
        "maxGain": 10,
        "reason": "No machine payment protocol (0). Per instance-hour prices for GPUs, CPUs, storage and networking are published without a login, with some models quoted by sales (20). No free tier or self-serve trial found. Trials run through the sales-led ARENA programme (0). An organisation has to be approved by the sales team and a person creates the token in the console (0).",
        "checklist": [
          "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.\n- 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.\n- 20, a free tier or trial that doesn't need a card.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-payments"
      },
      {
        "key": "ergonomics",
        "name": "Agent ergonomics",
        "score": 58,
        "maxGain": 6.8,
        "reason": "Graded on the platform API, the Kubernetes API and the Mission Control MCP server together. The MCP server documents 38 tools, most of them small read-only queries with limits, and every tool also takes an optional `user_prompt` (12 of 25). The Dedicated Inference API pages with `maxPageSize` and `pageToken`. The cluster list returns everything, with `returnPartialSuccess` for unreachable zones (14). Errors follow the google.rpc shape with a numeric code and a message, and support articles explain common failures (12). No idempotency keys, no MCP read-only or destructive annotations documented, and `coreweave_kubectl_apply` ignores `dry_run` and `plan_token` and applies the manifest (8). Generated clients on Buf for Go, Python, TypeScript, Java and others, plus a Terraform provider. Creating a cluster needs a zone, version and VPC (12).",
        "checklist": [
          "- 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).\n- 20, pagination, filtering and output-size controls.\n- 20, actionable, documented error responses, codes and messages an agent can recover from.\n- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-ergonomics"
      },
      {
        "key": "security",
        "name": "Security \u0026 auth",
        "score": 74,
        "maxGain": 4.6,
        "reason": "API access tokens are user-scoped, carry an expiry, are shown once and can be revoked in the console. They inherit the user's IAM roles and have no scopes of their own. OIDC workload identity federation is available for CKS workloads. Tokens travel in the Authorization header only (25). IAM has a Viewer and an Admin role per service, and the MCP write tools are off unless CoreWeave enables them for an organisation. There is no confirmation step, and the apply tool ignores `dry_run` (15). The MCP server returns log lines and documentation text, which are untrusted content, and no injection guidance was found (5). Kubernetes audit logs have a Grafana dashboard, Telemetry Relay forwards audit logs to HTTPS or S3, and support staff access is controlled and logged (15). Vulnerability disclosure policy with safe harbour, trust centre badges for SOC 2 Type II, ISO 27001, 27017 and 27018, and a security advisories page whose latest entry is July 2024. No security.txt and no bug bounty found (14).",
        "checklist": [
          "- 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.\n- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.\n- 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.\n- 0 to 15, audit logs or per-call visibility for the operator.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-security"
      },
      {
        "key": "schema",
        "name": "Schema \u0026 documentation",
        "score": 79,
        "maxGain": 3.4,
        "reason": "OpenAPI 3.0 specs for the CKS, VPC, Object Storage, Dedicated Inference, Capacity Finder and Telemetry Relay APIs are embedded in the reference pages, and the same services are published as Protobuf on the Buf Schema Registry. The raw .yaml paths in llms.txt redirected our reader to a docs login (25). llms.txt at docs.coreweave.com and every page served as Markdown (10). Operations and MCP tools state their purpose and inputs, with few lines on when not to use them (12). Required fields are marked and the Node Pool reference lists enums, but the CKS spec has none and instance types and versions are plain strings (9). Each operation page has a curl example. Errors are one generic `code`, `message`, `details` shape with no per-operation codes (8). Versioned paths and a dated changelog of 168 entries since December 2024 (15).",
        "checklist": [
          "APIs and MCP servers.",
          "- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).\n- 10, llms.txt or Markdown docs served for agents.\n- 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.\n- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.\n- 0 to 15, examples and documented error responses.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-schema"
      },
      {
        "key": "transparency",
        "name": "Transparency \u0026 trust",
        "score": 77,
        "maxGain": 2,
        "reason": "Closed service with published terms. They were last modified on 30 June 2022 and say they may be amended without notice (15). The terms commit to deleting customer data after account closure within a 30-day recovery period and at most 180 days, there is a data processing agreement, and the privacy policy of 24 February 2026 gives no fixed retention period for personal data (22). The changelog gives dated deprecations, for example operating system and driver versions from 29 July 2026, and the maintenance policy promises 24 hours' notice for critical work, though it can change without notice (14). Three sub-processors are named with their country and 30 days' notice of additions, and regions and availability zones are documented (20).",
        "blend": "editorial 71, provenance 83",
        "checklist": [
          "- 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.\n- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).\n- 0 to 20, a deprecation policy or notices with dates.\n- 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."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-transparency"
      },
      {
        "key": "maintenance",
        "name": "Maintenance \u0026 community",
        "score": 80,
        "maxGain": 1.8,
        "reason": "The changelog's latest entry is 2 October 2026 (30). It has 18 dated entries between 10 July and 2 October 2026 (20). Closed service with a dated changelog, a support centre and status updates by email. No public issue tracker for the platform (10 of 15). Generated clients on the Buf Schema Registry and a Terraform provider tagged v0.24.0 on 11 September 2026, the eighth tag since 14 July. The MCP server isn't in the official MCP registry (12). The provider repository has build, test, acceptance-test, release and Renovate workflows. We didn't check that they pass (8).",
        "checklist": [
          "- 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.\n- 20, at least three releases or dated changelog entries in the last 90 days.\n- 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.\n- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).\n- 10, package health, current dependencies and CI.",
          "Models are read for deprecation notice periods and model churn rather than release counts."
        ],
        "checklistUrl": "https://www.anchorterminal.com/benchmark/#checklist-maintenance"
      }
    ],
    "provenance": [
      {
        "label": "Domain age",
        "value": "coreweave.com, registered 2019-04-01 (7 years)",
        "points": 11,
        "max": 15
      },
      {
        "label": "Terms of service",
        "value": "read, states 6 of the 7 things a reader expects, and has 1 clause that costs points",
        "points": 7.1,
        "max": 10
      },
      {
        "label": "security.txt",
        "value": "not found",
        "points": 0,
        "max": 10
      }
    ],
    "unchecked": [
      "unchecked: status history before 25 September 2026. The Status.io history page is drawn by script and the feed keeps ten items, so the 90-day record is scored on two weeks.",
      "unchecked: the raw OpenAPI .yaml files named in llms.txt redirected to a docs login, so the specs were read as embedded in the reference pages.",
      "unchecked: whether the Terraform provider's CI passes on the default branch.",
      "unchecked: id.coreweave.com/signup is drawn by script. The docs say the sales team approves each organisation, and we couldn't see whether that page opens an infrastructure account without approval.",
      "unchecked: the trust centre's documents sit behind an access request, so the certifications are badges we read, not reports.",
      "No request rate limits, billing increment for on-demand nodes or bug bounty were found in the reviewed documentation.",
      "The lead lists Serverless Inference as part of this product. It is a Forge service at api.inference.wandb.ai with its own key and plans, so it is noted here and not graded."
    ],
    "weaknesses": [
      "No self-serve route. CoreWeave's sales team approves an organisation and emails the activation link before a token can be created",
      "No request rate limits, 429 guidance or idempotency keys found for api.coreweave.com in the reviewed documentation",
      "The CKS API is versioned v1beta1 and the Node Pool resource and Dedicated Inference API v1alpha1",
      "Three incidents between 26 September and 6 October 2026, including a US-CENTRAL-09A network outage of over two hours on 5 October",
      "The MCP tool `coreweave_kubectl_apply` ignores its `dry_run` parameter and applies the manifest, per the tool reference"
    ],
    "agentNotes": [
      "Create an API access token at console.coreweave.com/tokens with an expiry and send it as `Authorization: Bearer` to https://api.coreweave.com",
      "Add GPUs by applying a NodePool resource (`compute.coreweave.com/v1alpha1`) to a CKS cluster with `instanceType` and `targetNodes`. `instanceType` can't be changed afterwards",
      "Set `targetNodes` to 0 or delete the Node Pool when a job ends. Nodes are whole 8-GPU machines billed by the hour",
      "Never pass `dry_run` to the MCP tool `coreweave_kubectl_apply` expecting a preview. The reference says the parameter is ignored and the manifest is applied",
      "Use a separate Forge API key and https://api.inference.wandb.ai/v1 for Serverless Inference. A CoreWeave API access token doesn't work there"
    ],
    "recheck": "https://www.anchorterminal.com/builders/#disputes"
  },
  "meta": {
    "attribution": "Anchor Terminal (https://www.anchorterminal.com)",
    "docs": "https://www.anchorterminal.com/docs/",
    "generatedAt": "2026-10-08",
    "license": "CC-BY-4.0",
    "method": "https://www.anchorterminal.com/benchmark/",
    "methodology": "0.4",
    "openapi": "https://www.anchorterminal.com/openapi.json",
    "preview": false,
    "run": "2026-10-01",
    "runLabel": "October 2026 research run"
  }
}