# Fix list: Zulip From Anchor Terminal's listing at https://www.anchorterminal.com/tools/zulip, the October 2026 research run, assessed 9 October 2026. Grade C, 61.5 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 Zulip: work through the items below in the product, its docs and its public pages. Each category gives the reason for its score, with the points each checklist item earned, and the checklist itself, so the gap is the items that earned less than their points. Change the product, not the wording, and keep a note of what you changed and where it's published. ## 1. Security & auth, 42 out of 100, up to 10.1 more on the total Why it scored 42: One revocable API key per account with no scopes or expiry, and no OAuth for API clients, read as plain revocable keys (20). Incoming webhook URLs carry the key as `api_key` in the query string as a documented option (less 10). Bots are separate accounts, an incoming webhook bot can only send, roles and channel permissions apply to bots, and administrators can restrict bot creation. No read-only key and no confirmation for deletes (12). Messages are other users' text and no prompt-injection guidance was found (2). The security page claims an audit log of administrative actions. No help article or per-call log for an operator was found (5). A disclosure policy, a private HackerOne programme, advisories published with CVE numbers and CodeQL in CI. No security.txt, SOC 2 or ISO 27001 found (13). The checklist (https://www.anchorterminal.com/benchmark/#checklist-security): - 0 to 30, the credential model. 30 for OAuth 2.1 with scopes, or scoped and revocable keys with rotation. 20 for plain revocable API keys. 10 for one all-powerful key. 10 off when a secret can travel in a URL query string as a documented option. - 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions. - 0 to 15, prompt-injection posture where the tool returns untrusted content (documented mitigations or guidance). A tool that returns no untrusted content gets 10. - 0 to 15, audit logs or per-call visibility for the operator. - 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public. Models are read for retention, whether API data trains models (and whether that's off by default), zero-retention options and certifications. Frameworks for telemetry defaults, approval hooks, guardrails and sandboxing. ## 2. Payments & pricing, 30 out of 100, up to 8.8 more on the total Why it scored 30: No x402, MPP or L402 (0). Prices are public per user a month, $6.67 to $12 on the cloud plans, with nothing per call (10). The Free cloud plan needs no card (20). A person creates the organisation and the bot in a browser, and no programmatic route to a first key was found (0). The checklist (https://www.anchorterminal.com/benchmark/#checklist-payments): The published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/). - 40, a machine payment protocol (x402, MPP or L402) on the tool's own endpoints. 10 to 30 when it covers only some endpoints or only goes through a third party, and the note says which. - 20, per-call or per-unit pricing published without a login. 10 for public plan-only pricing, 0 for "contact sales" or prices behind a login. - 20, a free tier or trial that doesn't need a card. - 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API). Payment platforms and agent wallets rarely charge for their own API over a machine protocol, so the first line has steps for them, and the highest one that applies counts. 40 when x402, MPP or L402 runs on all their own endpoints, 30 when it runs on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, and 0 when the only protocol is their own. Merchant acceptance sits above a facilitator because the platform's own customers can charge agents through it, while a facilitator settles for sellers who wire up the protocol themselves. The counter-argument (a facilitator does more for the protocol as a whole) has a point. Each note says which step applied. Open-source software you run yourself is scored on its hosted or paid option if it has one. A free, self-hosted package with nothing to buy gets 20, 20 and 20 for the last three lines, and 0 to 40 for the first only if it ships a payment protocol. ## 3. Agent ergonomics, 68 out of 100, up to 5.2 more on the total Why it scored 68: `GET /messages` sizes responses with `num_before`, `num_after` and `apply_markdown`, and `POST /register` with `fetch_event_types`. No field selection (15). Anchor-based paging with flags for more results, and `narrow` filters using the search operators (18). Errors carry a stable `code` and extra keys such as `var_name`, with `msg` translated. Many failures share `BAD_REQUEST` (17). No idempotency keys. `local_id` and `queue_id` on `POST /messages` serve local echo, and the docs tell clients to repeat `delete-topic` until `complete` is true (6). Three required parameters to send a message, official Python and JavaScript libraries, the Python one last tagged in September 2025 (12). 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. ## 4. Reliability, 83 out of 100, up to 3.4 more on the total Why it scored 83: Graded on the hosted Zulip Cloud API with the hosted lines. Status page at status.zulip.com on Statuspage with nine components (20). The history lists no incidents from June to 8 October 2026 and one minor incident on 21 May. The Static asset CDN component was marked degraded on the day with no incident posted (30). The API docs give 200 requests a minute per user as the default configuration and say limits can vary by server and over time (13). A 429 returns `RATE_LIMIT_HIT` with `retry-after`, and every response has `X-RateLimit-*` headers. No idempotency key on `POST /messages` and no retry guidance for writes (10). No SLA found (0). The API is generally available (10). The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability): Hosted APIs, MCP servers, models and platforms. - 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own). - 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so. - 15, rate limits documented with numbers. - 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved. - 10, an SLA published for any paid tier. - 10, the surface agents use is generally available, not beta or preview. Local packages, SDKs, frameworks and stdio MCP servers. - 20, installs from an official package with supported runtimes stated. - 25, a public CI and test suite, passing on the default branch. - 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered). - 15, semver discipline and breaking changes called out in a changelog. - 15, version 1.0 or later, or declared stable. Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors. ## 5. Schema & documentation, 82 out of 100, up to 2.9 more on the total Why it scored 82: Public OpenAPI 3.0.1 file in the repository, 166 operations on 119 paths (25). No llms.txt. The docs are Markdown in the public repository and are served as HTML only (2). All 166 operations have a description, median 295 characters, many saying which clients an endpoint is for (16). 263 enums and 322 closed objects, but one `maxLength`, five minimums and several parameters that are JSON encoded inside form fields, such as `narrow` (11). 819 example values, curl, Python and JavaScript samples per endpoint, and a shared error page with codes (14). No version beyond `/api/v1`. Changes are tracked by feature level in a public API changelog with 499 entries and 975 inline change notes, none dated (14). The checklist (https://www.anchorterminal.com/benchmark/#checklist-schema): APIs and MCP servers. - 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool). - 10, llms.txt or Markdown docs served for agents. - 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference. - 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs. - 0 to 15, examples and documented error responses. - 15, versioning and a public changelog. Models are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference. ## 6. Transparency & trust, 76 out of 100, up to 2.1 more on the total Made of editorial 77, provenance 74. Why it scored 76: The server is open source under Apache 2.0 (30). Terms, a privacy policy, rules of use, a DPA and a GDPR page agree that Kandra Labs is processor for customer data. The privacy policy keeps personal data while an account is open and states no periods, and the DPA is a PDF we did not read (20). The release lifecycle keeps the API compatible with apps released in the last 12 months, and the changelog marks removals by feature level, 50 fields are marked deprecated, with no dates (13). Seven sub-processors are named with their roles, and the service is hosted in the United States. No country per sub-processor and no change log for the list (14). 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): - Endpoint on the vendor's domain: is not on zulip.com (0 of 15) - Terms of service: read, states 6 of the 7 things a reader expects (9.1 of 10) - security.txt: not found (0 of 10) ## 7. Maintenance & community, 77 out of 100, up to 2 more on the total Why it scored 77: Zulip Server 12.3 was released on 21 September 2026, 18 days before the check, and the main branch had a commit on 8 October (30). Two tagged releases since 11 July 2026, 12.2 and 12.3. The API changelog adds 13 undated feature levels for 13.0 and the cloud runs ahead of the tags, so half marks (10). The newest open issues were all updated between 1 and 8 October 2026, with about 1,200 issues and 874 pull requests open on a repository of 26,014 stars, and a public development community chat (20). Official Python and JavaScript libraries exist. The Python tag 0.9.1 dates from 30 September 2025 and the JavaScript repository's last commit from December 2025 (8). CI, CodeQL and zizmor workflows, and dependency updates in each maintenance release (9). The checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance): - 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older. - 20, at least three releases or dated changelog entries in the last 90 days. - 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15. - 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models). - 10, package health, current dependencies and CI. Models are read for deprecation notice periods and model churn rather than release counts. ## Deductions Each comes off the total. A fixed and documented problem counts for less at the next check. - 2026-08-10 and 2026-09-21. GHSA-5r8f-gq2h-fcgp let a guest receive new messages from public channels it was not subscribed to by registering an event queue, fixed in 12.2. GHSA-42mq-rxcr-wj72 let a member forge the sender of a group direct message through the API, fixed in 12.3. Both are reachable with an ordinary API key. Fixed and published, so 2 points (https://github.com/zulip/zulip/security/advisories/GHSA-5r8f-gq2h-fcgp, https://github.com/zulip/zulip/security/advisories/GHSA-42mq-rxcr-wj72). - 2026-02-05 to 2026-09-21. Eleven further advisories in twelve months across releases 11.5, 11.6, 12.0, 12.2 and 12.3, among them CVE-2026-25742 (attachments still public after web-public access was disabled), CVE-2026-40300 (edit history exposing original content in the API) and GHSA-xw9h-9rcm-hx4m (OpenID Connect ignoring `email_verified`). All fixed and published by the vendor, so 2 points (https://github.com/zulip/zulip/blob/main/docs/overview/changelog.md). ## 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 Data Processing Addendum, a PDF at zulip.com/static/images/policies/Zulip-Data-Processing-Addendum.pdf - unchecked: the PyPI project page for `zulip`, which needs JavaScript. The version and date come from the repository tag 0.9.1 - unchecked: the rate limits Zulip Cloud applies. The 200 a minute figure is the documented default configuration - unchecked: whether a bot counts as a paid user on Standard and Plus - unchecked: the status API and uptime figures. status.zulip.com disallows `/api/` in robots.txt, so the history pages were read instead - unchecked: severity ratings of the advisories, and whether a second page of advisories lists more than the changelog does - The lead gave docs.zulip.com/api/rest. The API documentation read is at zulip.com/api/rest, and docs.zulip.com was not requested - The lead named the vendor Zulip. The legal entity in the terms is Kandra Labs, Inc. - No official MCP server was found in the help centre, API docs or server documentation - No SLA, SOC 2 or ISO 27001 statement was found on the pages read. One may exist under a sales agreement - The Rules of Use say not to permit any third party to access an account's credentials. How that applies to an agent run by a third party was not established - The audit log named on the security page has no help article we could find, so what an organisation administrator can see on Zulip Cloud was not established ## Weaknesses - One API key per account, with no scopes, expiry or OAuth. A generic bot's key does what a normal member can do - Thirteen security advisories between February and September 2026, among them guests reading unsubscribed public channels and forged senders through the API - `POST /messages` has no idempotency key, so a retried send can post twice - Incoming webhook URLs carry the bot's key in the query string as `api_key` - No SLA, SOC 2 or ISO 27001 statement, security.txt or llms.txt was found ## 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. - Ask the organisation for a bot of the most limited type that fits. Use an incoming webhook bot when the task only posts messages - Authenticate with HTTP Basic, the bot's email as user and its API key as password, against https://.zulipchat.com/api/v1 - Read `code`, not `msg`, on errors. `msg` is translated into the account's language - On `RATE_LIMIT_HIT` wait the seconds in `retry-after`. The default limit is 200 requests a minute per user - Fetch history with `GET /messages`, a `narrow` filter, an `anchor` and `num_before` or `num_after`, at most 1,000 a batch as the docs recommend ## 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.