{
  "fixes": {
    "slug": "braze",
    "name": "Braze",
    "listing": "https://www.anchorterminal.com/tools/braze",
    "markdown": "# Fix list: Braze\n\nFrom Anchor Terminal's listing at https://www.anchorterminal.com/tools/braze, the October 2026 research run, assessed 8 October 2026. Grade C, 61.2 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 Braze: 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. Payments \u0026 pricing, 10 out of 100, up to 11.3 more on the total\n\nWhy it scored 10: Read with the hosted rubric. No x402, MPP or L402 (0). The pricing page names four editions (Go, Select, Pro, Enterprise) priced by monthly active users and Action Credits, shows no figures and sends buyers to sales (0). It says there is no free plan. A 14-day trial is open at try.braze.com, and we couldn't read whether the form asks for a card, so half marks (10 of 20). A person signs up, creates a REST key in the dashboard or approves OAuth in a browser, and an admin must switch MCP access on first (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## 2. Agent ergonomics, 50 out of 100, up to 8.1 more on the total\n\nWhy it scored 50: 71 MCP tools is over the 30 line, with no toolsets or dynamic loading documented. What a user sees follows dashboard permissions and three OAuth scopes. On REST, exports take `fields_to_export` (10 of 25). List endpoints page by `page` in groups of 100 with `sort_direction` and `last_edit.time[gt]`, custom attributes return in groups of 50, and segment exports go to a file with an optional callback (15 of 20). Errors are documented as HTTP status plus a message string and an `errors` array for non-fatal problems, without stable codes. A success response means queued, and `/users/identify` returns success when nothing matched (14 of 20). No idempotency keys. `/users/track/status` reports on asynchronous requests tagged with a `group_id`, and the docs suggest waiting up to five minutes between dependent calls. We couldn't read MCP annotations (5 of 20). Few required parameters on reads. The only REST client library is a Ruby one marked beta, last pushed in January 2022 (6 of 15).\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## 3. Reliability, 70 out of 100, up to 6 more on the total\n\nWhy it scored 70: Read with the hosted lines and scored on the REST API first, with the remote MCP server beside it. status.braze.com on Statuspage has 132 components across 15 clusters, with a REST APIs component for each cluster (20). Between 10 July and 8 October 2026 it lists seven entries. One reached the API, a minor incident on 5 August with elevated error rates on US-01 and US-03 from 11:30 to 12:55 ET. The one critical entry, on 16 September, was the support portal during a Salesforce outage, and the rest were dashboard latency on US-07, Currents delay on EU-02 and Operator errors (18 of 30, minor incidents only, less 2 for 85 minutes of API errors on two clusters). REST limits are published per endpoint, 250,000 requests an hour by default and 250 a minute for broadcast sends. None were found for the MCP server (13 of 15). Every response carries `X-RateLimit-Limit`, `-Remaining` and `-Reset`, and the docs advise exponential backoff on 5XX. There's no Retry-After, no idempotency key for sends, and the errors page says Braze can't replay REST calls that fail during an incident (9 of 15). No published SLA. The terms promise commercially reasonable efforts and leave service levels to the contract (0). The REST API is generally available and the MCP server has been since 17 September 2026, with 21 of its 71 tools limited to beta programmes (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## 4. Schema \u0026 documentation, 68 out of 100, up to 5.2 more on the total\n\nWhy it scored 68: No OpenAPI document was found in the docs or the braze-inc GitHub organisation. A public Postman collection covers the REST endpoints, and the MCP tool schemas need a Braze sign-in that we don't have (12 of 25). www.braze.com/llms.txt exists for the marketing site, every docs page has a Markdown copy at /index.md and a Copy for LLM button, and the docs are on Context7 (10). Each endpoint page states its purpose, the key permission it needs and its rate limit. The MCP page lists 71 tools with a one-line description, the REST endpoint behind each and its access type, and says when to use the REST API instead of Operator Connect (15 of 20). Parameters come in tables with required flags and data types, without a schema to validate against (9 of 15). Curl and response examples on each endpoint page, and a table of fatal errors with causes (13 of 15). Release notes are dated and monthly. The REST API has no version scheme, so there is nothing for a client to pin (9 of 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## 5. Security \u0026 auth, 77 out of 100, up to 4 more on the total\n\nWhy it scored 77: The MCP server uses OAuth with PKCE S256, dynamic client registration, short-lived access tokens with refresh and the scopes `mcp:tools`, `mcp:resources` and `mcp:operator`, checked against an allowlist of redirect URIs. REST keys are limited to named endpoint permissions and optional IP allowlists and travel in a Bearer header. MCP tokens can be revoked only for the whole company at once (29 of 30). MCP access mirrors the user's dashboard permissions, needs a Use MCP Server permission that is off by default in each workspace, and has no tools that return user-level personal data. There is no confirmation step, Operator Connect acts without per-step approval, and Braze advises against a client's auto mode (15 of 20). Campaign content, catalogue items and attribute names can carry text written by others. The docs tell users to review proposed actions and give no prompt-injection guidance (5 of 15). The security event report records OAuth connections and REST key creation and removal, with export to S3, and an API usage dashboard shows traffic. Edits by individual MCP update tools don't appear in campaign edit history (9 of 15). security.txt valid to 1 January 2031 pointing to a HackerOne programme, ISO 27001, SOC 2 Type II and an annual third-party penetration test. We couldn't read whether the programme pays bounties (19 of 20).\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## 6. Maintenance \u0026 community, 69 out of 100, up to 2.7 more on the total\n\nWhy it scored 69: The newest release notes are dated 17 September 2026 and move the remote MCP server to general availability (30). Three dated releases since 10 July (23 July, 20 August and 17 September), which just meets the line (20). Closed service with public release notes, a support portal, a product feedback route for MCP clients and the Bonfire community. We didn't test support (10 of 15). The server isn't in the official MCP registry, where a search for braze returns only a third-party project. There is no current official REST SDK, and the mobile and web SDK repositories were pushed in October 2026 (4 of 15). The earlier local MCP server on PyPI, 1.2.0 of 14 May 2026, is deprecated with no further updates (5 of 10).\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## 7. Transparency \u0026 trust, 83 out of 100, up to 1.5 more on the total\n\nMade of editorial 69, provenance 96.\n\nWhy it scored 83: Closed service under a Main Subscription Agreement last updated 13 October 2025, with SDK source public (15). The data retention page keeps profile data for the life of the account, up to 90 days in internal systems, two years before personal data is removed from the Data Lake and six months in backups. The DPA (revision 28 August 2025) commits to export, deletion and a certificate of deletion at termination, and the privacy policy is dated 23 June 2026. The retention page was last revised on 1 April 2024 (25 of 30). A deprecations log with dates exists and the local MCP server is marked deprecated, but no notice period for REST endpoints or MCP tools was found (9 of 20). The sub-processor list of 1 June 2026 gives each entity's service and location with 30 days' notice of additions, and the security datasheet names the AWS regions for US, EU, Australia, Indonesia, Japan and South Korea hosting (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- Terms of service: read, states 5 of the 7 things a reader expects, and has 1 clause that costs points (6.3 of 10)\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: whether the 14-day trial at try.braze.com asks for a card. The form is rendered by JavaScript and we couldn't read it\n- unchecked: MCP tool input schemas, annotations and error shapes, which need a Braze sign-in\n- unchecked: whether the HackerOne programme named in security.txt pays bounties. The page didn't render for our reader\n- unchecked: weekly downloads of braze-mcp-server on PyPI. pypistats.org answered 429\n- No OpenAPI document, published SLA, MCP rate limit or REST deprecation notice period was found in the reviewed documentation\n- The trial page says 14 days and the free trial terms say 30 days unless otherwise specified. We recorded 14\n- Scored on the REST API first because the MCP server has no profile, event or send tools. A reader who needs only analytics and content would weigh the MCP lines more\n\n## Weaknesses\n\n- No public prices. Editions, monthly active users and Action Credits are sold by contract, and the pricing page says there is no free plan\n- No OpenAPI document found. The machine-readable contract is a public Postman collection\n- No idempotency keys on sends or `/users/track`, and Braze says it can't replay REST calls that fail during an incident\n- 71 MCP tools with no documented toolsets, 21 of them limited to beta programmes\n- Edits made through individual MCP update tools don't appear in campaign edit history, and Operator Connect acts without per-step approval\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- Use the REST host for the customer's instance, such as https://rest.iad-01.braze.com or https://rest.fra-01.braze.eu. A key sent to the wrong instance returns 401\n- Create a REST key with only the permissions the task needs. Scope and IP allowlist can't be edited later, so a change means a new key\n- Call `get_workspaces` first on MCP. Every other tool needs the returned workspace id as `app_group_id`, and the wrong workspace is a documented failure\n- Use REST for profiles, events and sends. The MCP server has no user-level tools and can't trigger a message to a named user\n- Treat `\"message\": \"success\"` as queued, not delivered. Add a `group_id` to `/users/track` and poll `/users/track/status`, and don't resend a trigger call blindly\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": 61.2,
    "assessed": "2026-10-08",
    "run": "October 2026 research run",
    "categories": [
      {
        "key": "payments",
        "name": "Payments \u0026 pricing",
        "score": 10,
        "maxGain": 11.3,
        "reason": "Read with the hosted rubric. No x402, MPP or L402 (0). The pricing page names four editions (Go, Select, Pro, Enterprise) priced by monthly active users and Action Credits, shows no figures and sends buyers to sales (0). It says there is no free plan. A 14-day trial is open at try.braze.com, and we couldn't read whether the form asks for a card, so half marks (10 of 20). A person signs up, creates a REST key in the dashboard or approves OAuth in a browser, and an admin must switch MCP access on first (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": 50,
        "maxGain": 8.1,
        "reason": "71 MCP tools is over the 30 line, with no toolsets or dynamic loading documented. What a user sees follows dashboard permissions and three OAuth scopes. On REST, exports take `fields_to_export` (10 of 25). List endpoints page by `page` in groups of 100 with `sort_direction` and `last_edit.time[gt]`, custom attributes return in groups of 50, and segment exports go to a file with an optional callback (15 of 20). Errors are documented as HTTP status plus a message string and an `errors` array for non-fatal problems, without stable codes. A success response means queued, and `/users/identify` returns success when nothing matched (14 of 20). No idempotency keys. `/users/track/status` reports on asynchronous requests tagged with a `group_id`, and the docs suggest waiting up to five minutes between dependent calls. We couldn't read MCP annotations (5 of 20). Few required parameters on reads. The only REST client library is a Ruby one marked beta, last pushed in January 2022 (6 of 15).",
        "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": "reliability",
        "name": "Reliability",
        "score": 70,
        "maxGain": 6,
        "reason": "Read with the hosted lines and scored on the REST API first, with the remote MCP server beside it. status.braze.com on Statuspage has 132 components across 15 clusters, with a REST APIs component for each cluster (20). Between 10 July and 8 October 2026 it lists seven entries. One reached the API, a minor incident on 5 August with elevated error rates on US-01 and US-03 from 11:30 to 12:55 ET. The one critical entry, on 16 September, was the support portal during a Salesforce outage, and the rest were dashboard latency on US-07, Currents delay on EU-02 and Operator errors (18 of 30, minor incidents only, less 2 for 85 minutes of API errors on two clusters). REST limits are published per endpoint, 250,000 requests an hour by default and 250 a minute for broadcast sends. None were found for the MCP server (13 of 15). Every response carries `X-RateLimit-Limit`, `-Remaining` and `-Reset`, and the docs advise exponential backoff on 5XX. There's no Retry-After, no idempotency key for sends, and the errors page says Braze can't replay REST calls that fail during an incident (9 of 15). No published SLA. The terms promise commercially reasonable efforts and leave service levels to the contract (0). The REST API is generally available and the MCP server has been since 17 September 2026, with 21 of its 71 tools limited to beta programmes (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": "schema",
        "name": "Schema \u0026 documentation",
        "score": 68,
        "maxGain": 5.2,
        "reason": "No OpenAPI document was found in the docs or the braze-inc GitHub organisation. A public Postman collection covers the REST endpoints, and the MCP tool schemas need a Braze sign-in that we don't have (12 of 25). www.braze.com/llms.txt exists for the marketing site, every docs page has a Markdown copy at /index.md and a Copy for LLM button, and the docs are on Context7 (10). Each endpoint page states its purpose, the key permission it needs and its rate limit. The MCP page lists 71 tools with a one-line description, the REST endpoint behind each and its access type, and says when to use the REST API instead of Operator Connect (15 of 20). Parameters come in tables with required flags and data types, without a schema to validate against (9 of 15). Curl and response examples on each endpoint page, and a table of fatal errors with causes (13 of 15). Release notes are dated and monthly. The REST API has no version scheme, so there is nothing for a client to pin (9 of 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": "security",
        "name": "Security \u0026 auth",
        "score": 77,
        "maxGain": 4,
        "reason": "The MCP server uses OAuth with PKCE S256, dynamic client registration, short-lived access tokens with refresh and the scopes `mcp:tools`, `mcp:resources` and `mcp:operator`, checked against an allowlist of redirect URIs. REST keys are limited to named endpoint permissions and optional IP allowlists and travel in a Bearer header. MCP tokens can be revoked only for the whole company at once (29 of 30). MCP access mirrors the user's dashboard permissions, needs a Use MCP Server permission that is off by default in each workspace, and has no tools that return user-level personal data. There is no confirmation step, Operator Connect acts without per-step approval, and Braze advises against a client's auto mode (15 of 20). Campaign content, catalogue items and attribute names can carry text written by others. The docs tell users to review proposed actions and give no prompt-injection guidance (5 of 15). The security event report records OAuth connections and REST key creation and removal, with export to S3, and an API usage dashboard shows traffic. Edits by individual MCP update tools don't appear in campaign edit history (9 of 15). security.txt valid to 1 January 2031 pointing to a HackerOne programme, ISO 27001, SOC 2 Type II and an annual third-party penetration test. We couldn't read whether the programme pays bounties (19 of 20).",
        "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": "maintenance",
        "name": "Maintenance \u0026 community",
        "score": 69,
        "maxGain": 2.7,
        "reason": "The newest release notes are dated 17 September 2026 and move the remote MCP server to general availability (30). Three dated releases since 10 July (23 July, 20 August and 17 September), which just meets the line (20). Closed service with public release notes, a support portal, a product feedback route for MCP clients and the Bonfire community. We didn't test support (10 of 15). The server isn't in the official MCP registry, where a search for braze returns only a third-party project. There is no current official REST SDK, and the mobile and web SDK repositories were pushed in October 2026 (4 of 15). The earlier local MCP server on PyPI, 1.2.0 of 14 May 2026, is deprecated with no further updates (5 of 10).",
        "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"
      },
      {
        "key": "transparency",
        "name": "Transparency \u0026 trust",
        "score": 83,
        "maxGain": 1.5,
        "reason": "Closed service under a Main Subscription Agreement last updated 13 October 2025, with SDK source public (15). The data retention page keeps profile data for the life of the account, up to 90 days in internal systems, two years before personal data is removed from the Data Lake and six months in backups. The DPA (revision 28 August 2025) commits to export, deletion and a certificate of deletion at termination, and the privacy policy is dated 23 June 2026. The retention page was last revised on 1 April 2024 (25 of 30). A deprecations log with dates exists and the local MCP server is marked deprecated, but no notice period for REST endpoints or MCP tools was found (9 of 20). The sub-processor list of 1 June 2026 gives each entity's service and location with 30 days' notice of additions, and the security datasheet names the AWS regions for US, EU, Australia, Indonesia, Japan and South Korea hosting (20).",
        "blend": "editorial 69, provenance 96",
        "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"
      }
    ],
    "provenance": [
      {
        "label": "Terms of service",
        "value": "read, states 5 of the 7 things a reader expects, and has 1 clause that costs points",
        "points": 6.3,
        "max": 10
      }
    ],
    "unchecked": [
      "unchecked: whether the 14-day trial at try.braze.com asks for a card. The form is rendered by JavaScript and we couldn't read it",
      "unchecked: MCP tool input schemas, annotations and error shapes, which need a Braze sign-in",
      "unchecked: whether the HackerOne programme named in security.txt pays bounties. The page didn't render for our reader",
      "unchecked: weekly downloads of braze-mcp-server on PyPI. pypistats.org answered 429",
      "No OpenAPI document, published SLA, MCP rate limit or REST deprecation notice period was found in the reviewed documentation",
      "The trial page says 14 days and the free trial terms say 30 days unless otherwise specified. We recorded 14",
      "Scored on the REST API first because the MCP server has no profile, event or send tools. A reader who needs only analytics and content would weigh the MCP lines more"
    ],
    "weaknesses": [
      "No public prices. Editions, monthly active users and Action Credits are sold by contract, and the pricing page says there is no free plan",
      "No OpenAPI document found. The machine-readable contract is a public Postman collection",
      "No idempotency keys on sends or `/users/track`, and Braze says it can't replay REST calls that fail during an incident",
      "71 MCP tools with no documented toolsets, 21 of them limited to beta programmes",
      "Edits made through individual MCP update tools don't appear in campaign edit history, and Operator Connect acts without per-step approval"
    ],
    "agentNotes": [
      "Use the REST host for the customer's instance, such as https://rest.iad-01.braze.com or https://rest.fra-01.braze.eu. A key sent to the wrong instance returns 401",
      "Create a REST key with only the permissions the task needs. Scope and IP allowlist can't be edited later, so a change means a new key",
      "Call `get_workspaces` first on MCP. Every other tool needs the returned workspace id as `app_group_id`, and the wrong workspace is a documented failure",
      "Use REST for profiles, events and sends. The MCP server has no user-level tools and can't trigger a message to a named user",
      "Treat `\"message\": \"success\"` as queued, not delivered. Add a `group_id` to `/users/track` and poll `/users/track/status`, and don't resend a trigger call blindly"
    ],
    "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"
  }
}