Kroki

# Fix list: Kroki

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/kroki, the October 2026 research run, assessed 9 October 2026. Grade C, 59.2 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 Kroki: 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. Schema & documentation, 48 out of 100, up to 8.5 more on the total

Why it scored 48: No OpenAPI file or other machine-readable contract. `GET /health` returns the diagram types and library versions as JSON (3). No llms.txt on kroki.io or docs.kroki.io (HTTP 404). The docs are AsciiDoc files in the repository (2). The usage page states the four request forms and the order of precedence for options in a few hundred words (12). Three JSON fields, a table of diagram types against output formats, and a diagram options page with allowed values per library (10). Examples for cURL and HTTPie, and encoding samples in ten languages. Status codes 400, 408, 413, 414, 422 and 431 are described where each limit is documented, with no error catalogue, and the JSON error shape is not in the docs (9). A public changelog with dated versions. Paths carry no API version apart from a `/v1/health` alias (12).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-schema):

APIs and MCP servers.

- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).
- 10, llms.txt or Markdown docs served for agents.
- 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference.
- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.
- 0 to 15, examples and documented error responses.
- 15, versioning and a public changelog.

Models are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference.

## 2. Security & auth, 56 out of 100, up to 7.7 more on the total

Why it scored 56: No credential exists to leak or scope. The server has no authentication on its convert endpoints and binds all interfaces by default, so access control is the owner's network (18). `KROKI_SAFE_MODE` defaults to `SECURE`, which blocks file and network reads by diagram libraries, the container runs as the non-root user `kroki`, and the registration API is off in `SECURE` mode. Four advisories in 2026 showed ways round `SECURE` (13). The server renders the caller's own text and returns an image, with remote includes off by default (9). Request logs, `GET /metrics` and optional OpenTelemetry tracing (6). `SECURITY.md` asks for private GitHub advisories, four advisories were published with fixes in July and August 2026, a Trivy scan of the public images runs nightly and Renovate runs weekly. No security.txt (HTTP 404), bug bounty or certification (10).

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.

## 3. Reliability, 74 out of 100, up to 5.2 more on the total

Why it scored 74: Read with the local-software lines, because the listing grades the server an owner runs. The free public instance has no status page, SLA or published limit. Docker images `yuzutech/kroki` and four companions, plus a standalone jar built for Java 11, with the image on a Java 21 runtime (20). The `main.yaml` workflow passed on the last ten pushes to `main`, the latest on 7 October 2026 (25). GitHub's search counts 136 open issues. Most of the 30 newest are proposals for new diagram types and nearly all have a reply, while a September 2025 report of Mermaid PNG faults under high concurrency has none (17). CHANGELOG.md follows Keep a Changelog and states semantic versioning, and 0.33.0 records its change of the default Mermaid layout under Changed with the option that restores the old one (12). Version 0.33.0, with no 1.0 and no statement of stability, and diagrams.net marked experimental (0).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):

Hosted APIs, MCP servers, models and platforms.

- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).
- 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so.
- 15, rate limits documented with numbers.
- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.
- 10, an SLA published for any paid tier.
- 10, the surface agents use is generally available, not beta or preview.

Local packages, SDKs, frameworks and stdio MCP servers.

- 20, installs from an official package with supported runtimes stated.
- 25, a public CI and test suite, passing on the default branch.
- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).
- 15, semver discipline and breaking changes called out in a changelog.
- 15, version 1.0 or later, or declared stable.

Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors.

## 4. Payments & pricing, 60 out of 100, up to 5 more on the total

Why it scored 60: Read with the self-hosted rule. Free software with nothing to buy and no payment protocol (0). No price to publish (20). No card or trial needed (20). An agent can pull the Docker image, or call the public instance, with no signup (20).

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.

## 5. Agent ergonomics, 70 out of 100, up to 4.9 more on the total

Why it scored 70: One request with diagram text in and an image out, and no tool definitions to load. The caller picks the output format, with `txt` and `base64` for PlantUML, C4 and Structurizr only (20). No pagination is needed. Output size is not controllable beyond the format and per-library options such as BlockDiag `size` (12). Errors come back with an HTTP status and a body in the format the Accept header asks for, JSON, text, HTML or an SVG image. This is in `ErrorHandler.java` and not in the docs (12). Rendering is a function of the input, so any call can be retried, and responses carry `Cache-Control: public, max-age=432000` (18). One required field per request. No official SDKs. The Go CLI's newest release is v0.5.0 of December 2022 (8).

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.

## 6. Transparency & trust, 62 out of 100, up to 3.3 more on the total

Made of editorial 67, provenance 56.

Why it scored 62: MIT licence, with every component in one public repository (30). No privacy policy or terms for the software or for the public instance. The docs call the server stateless and say the public instance makes no remote calls. The kroki.io home page loads a Carbon Ads script (15). `KROKI_PORT` is marked deprecated with the reason and a replacement, and no removal date. Removed behaviour is recorded in the changelog (8). OpenTelemetry tracing is configured by environment variable. The 0.32.0 changelog discloses that the companions' headless Chromium contacted Google hosts at launch and that the calls are now blocked (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):

- Domain age: kroki.io, registered 2019-01-06 (7 years) (11 of 15)
- Terms of service: not found (0 of 10)
- Privacy policy: not found (0 of 10)
- Status page: not found (0 of 10)
- security.txt: not found (0 of 10)

## 7. Maintenance & community, 86 out of 100, up to 1.2 more on the total

Why it scored 86: Version 0.33.0 on 5 October 2026, four days before the check (30). Five versions in the last 90 days, 0.31.1 on 15 July, 0.31.2, 0.32.0, 0.32.1 and 0.33.0 (20). 97 commits since 11 July 2026, replies on nearly all of the 30 newest issues, and each 2026 advisory published within a day of the version that fixed it (20). The public instance reported 0.33.0 four days after the tag. No official SDKs, and the CLI has had no release since December 2022 (7). Renovate batches dependency updates weekly with a seven-day minimum age, CI runs on every push and a CVE scan runs nightly (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.

- 27 July to 12 August 2026. Four advisories on the repository. GHSA-wmpp-fj9c-w766 (critical, CVSS 9.8) allowed unauthenticated remote code execution on `/tikz/svg` in 0.21.0 up to 0.32.0 whatever the safe mode. GHSA-r54f-fq6c-53vw (high, CVE-2026-102359), GHSA-px99-rjv4-49g8 (medium, CVE-2026-102356) and GHSA-9p7m-vrmg-qp4q (high) let TikZ, Mermaid and Vega diagrams read local files or fetch URLs in `SECURE` mode. All are fixed, in 0.32.1 at the latest, and the maintainers published each with a changelog entry, so the deduction is five points (https://github.com/yuzutech/kroki/security/advisories).

## 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.

- No terms or privacy document was found for the software or the public instance, so `provenance.terms` and `provenance.privacy` are left out
- The legal form of Yuzu tech. Its site names none, and no company register was searched
- What the public instance logs or keeps. Nothing is published, and the listing grades the server an owner runs
- Whether OpenTelemetry sends anything when no `OTEL_*` variable is set. `Main.java` initialises the autoconfigured SDK at start and the exporter default was not traced
- How many of the 136 open issues are crashes or regressions. Only the 30 newest were read
- unchecked: Docker Hub pull counts and image tags. hub.docker.com was not asked
- unchecked: whether `https://demo.kroki.io`, the CLI's default endpoint, still answers. It was not called
- unchecked: the third-party hosts named in the README (Hostinger, Wovalab, Easypanel, ZopDay), which the project says it does not operate
- The gateway reads `KROKI_BODY_LIMIT` in `Server.java`, and the docs name only `KROKI_MAX_BODY_SIZE` for the companions

## Weaknesses

- GHSA-wmpp-fj9c-w766 (CVSS 9.8) allowed unauthenticated remote code execution on `/tikz/svg` in versions 0.21.0 up to 0.32.0, whatever the safe mode. Fixed 3 August 2026
- Three more advisories in July and August 2026 let TikZ, Mermaid and Vega diagrams read local files or fetch URLs in `SECURE` mode. All are fixed in 0.32.1
- No OpenAPI file, llms.txt or error catalogue. The JSON error shape is in the source and not in the documentation
- The public instance at kroki.io has no terms, privacy policy, status page or published rate limit. The CLI page limits the demonstration server to reasonable, non-commercial use
- The server has no authentication on its convert endpoints and binds all interfaces on port 8000 by default. The version is 0.33.0, with no 1.0

## What costs an agent a turn today

The notes we give agents before they call it. Each one is a workaround an agent shouldn't need.

- Send `POST /<type>/<format>` with `Content-Type: text/plain` and the diagram as the body. This avoids the deflate and base64 encoding that GET needs
- Send `Accept: application/json` on a JSON request to get errors as `{"error": {"code", "message"}}`. With an SVG Accept header the error arrives as an image
- Call `GET /health` first to list the diagram types and library versions the instance has. Mermaid, BPMN, Excalidraw and diagrams.net need companion containers
- Run 0.32.1 or later before rendering untrusted text. Earlier versions allow remote code execution through `/tikz/svg` and file reads in `SECURE` mode
- Self-host for private diagrams with `docker run -p8000:8000 yuzutech/kroki`, and set `KROKI_LISTEN=127.0.0.1:8000` or a network rule, since the server has no authentication

## 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.