agentic-mermaid 0.1.0 → 0.1.1

package/AGENT_NATIVE.md

@@ -146,6 +146,7 @@ Tier 3 warnings are family-specific quality hints for "common LLM mistakes" that
 | `COMMENT_DROPPED` | warning | The source contains in-body `%%` comments that this diagram's structured serialization does not preserve (reported with `count` and `lines`). The leading wrapper — frontmatter, `%%{init}%%` directives, comments before the header — always round-trips byte-verbatim; in-body comments survive only in opaque bodies or preserved opaque segments. Re-home load-bearing comments into the wrapper, or accept the loss as canonicalization. |
 | `UNSUPPORTED_SYNTAX` | warning | The source uses Mermaid syntax that is preserved losslessly but not fully modeled by local structured mutation/render semantics (for example flowchart edge IDs, edge metadata, click/href directives, or markdown strings). Payload includes `syntax`, optional `line`, and `message`. |
 | `CONTENT_DROPPED_ON_ROUNDTRIP` | warning | The structured `{nodes, edges, groups}` tally changed across a parse → serialize → re-parse cycle, so canonical serialization is silently dropping or duplicating content even though the bytes may re-parse (payload carries `before`/`after` counts). The faithfulness analogue of `COMMENT_DROPPED` — "100% parse success is not faithfulness". Runs on every verify, for every family; opaque bodies (byte-verbatim) are exempt. |
+| `INEFFECTIVE_CONFIG` | warning | A Mermaid config field was accepted (for config-shape compatibility) but has no effect on this family's geometry or paint — e.g. Journey's sequence-era fields (`boxMargin`, `rightAngles`, …). Payload names the `field`. Accepting-and-ignoring silently misleads migrating users; this lint says so. |
 
 `FamilyPlugin.verify` hooks are wired and run today; built-ins use them for Tier 1 structural warnings for class/ER and the central flowchart verifier emits the initial Tier 3 lint catalogue. Future lint codes should be added deliberately to `WARNING_TIER`, documented here, and covered by doc-sync tests.
 
@@ -295,20 +296,24 @@ Two contracts:
 | `add_relation`      | `from`, `to`, `leftCard`, `rightCard` (+ `dashed`, `label`) | `remove_relation(index)` |
 | `remove_relation`   | `index`                                               | `add_relation(...)` |
 
-**Journey MutationOp kinds** (10, BUILD-15 — the pilot promotion from opaque-only fallback semantics to structured mutation via the FamilyPlugin registry):
+**Journey MutationOp kinds** (14, BUILD-15 — the pilot promotion from opaque-only fallback semantics to structured mutation via the FamilyPlugin registry; ordering + accessibility ops added with the Mermaid-classic renderer elevation, since journey order IS the timeline):
 
 | Kind | Required | Inverse |
 |---|---|---|
 | `set_title`          | `title \| null`                                          | `set_title(prev_title)` |
-| `add_section`        | `label`                                                   | `remove_section(index)` |
+| `add_section`        | `label` (+ optional insert `index`)                       | `remove_section(index)` |
 | `remove_section`     | `index`                                                   | `add_section(label)` |
 | `set_section_label`  | `index`, `label`                                          | `set_section_label(index, prev_label)` |
-| `add_task`           | `sectionIndex`, `text`, `score` (+ optional `actors`)     | `remove_task(sectionIndex, taskIndex)` |
+| `add_task`           | `sectionIndex`, `text`, `score` (+ optional `actors`, insert `index`) | `remove_task(sectionIndex, taskIndex)` |
 | `remove_task`        | `sectionIndex`, `taskIndex`                               | `add_task(...)` |
 | `set_task_text`      | `sectionIndex`, `taskIndex`, `text`                       | `set_task_text(... prev_text)` |
 | `set_task_score`     | `sectionIndex`, `taskIndex`, `score` (integer 1..5)       | `set_task_score(... prev_score)` |
 | `set_task_actors`    | `sectionIndex`, `taskIndex`, `actors: string[]`           | `set_task_actors(... prev_actors)` |
 | `rename_actor`       | `from`, `to`                                              | `rename_actor(to, from)` |
+| `move_task`          | `fromSection`, `fromIndex`, `toSection`, `toIndex`        | `move_task(toSection, toIndex, fromSection, fromIndex)` |
+| `move_section`       | `from`, `to`                                              | `move_section(to, from)` |
+| `set_accessibility_title`       | `title \| null`                               | `set_accessibility_title(prev)` |
+| `set_accessibility_description` | `description \| null`                         | `set_accessibility_description(prev)` |
 
 **Architecture MutationOp kinds** (10, BUILD-17 — promoting the architecture-beta family to structured mutation via the FamilyPlugin registry, following the BUILD-15 journey pilot):
 

package/CHANGELOG.md

@@ -6,6 +6,11 @@ This changelog tracks user-facing changes for **Agentic Mermaid**, a fork of `lu
 
 _No changes yet._
 
+## 0.1.1 — 2026-07-10
+
+### Added
+- **Standards-based agent discovery and MCP Registry publishing.** Agent guidance now states when to use Agentic Mermaid and when to keep diagrams local. The website publishes `/index.md`, negotiates it from `/` with `Vary: Accept`, and advertises machine-readable resources with RFC 8288 links. The npm package ships official `server.json` metadata, exposes the registry launch path as `npx -y agentic-mermaid mcp`, and publishes each new release to the MCP Registry after npm succeeds.
+
 ## 0.1.0 — 2026-07-09
 
 ### Added

package/Instructions_for_agents.md

@@ -2,6 +2,12 @@
 
 This is the canonical agent-use guide. The same content is emitted by `am --agent-instructions`. A doc-sync test asserts the two are byte-identical.
 
+## When to use Agentic Mermaid
+
+Use Agentic Mermaid when an agent needs to create, edit, verify, describe, or render Mermaid diagrams through deterministic library, CLI, or MCP calls. It is strongest when the workflow needs typed edits, semantic fact read-back, CI-safe verification, or reproducible SVG, PNG, ASCII, and Unicode output.
+
+Do not use the hosted MCP for private diagrams; use the local library, CLI, or self-hosted MCP so source stays in your environment. Use another tool when the artifact is not Mermaid source or requires freeform canvas editing that Mermaid cannot express.
+
 ## Quick start
 
 Choose the narrowest channel. New diagrams: build with `buildMermaid(kind, ops)` — or `createMermaid(kind)` then typed mutations — and verify/render the result; author Mermaid source directly only for syntax the typed ops do not model. Existing structured diagrams: parse → narrow → mutate → verify → serialize. Code Mode/library are best for multi-step edits; CLI is best for one-shot verify/render/preview. Code Mode exposes `mermaid.*`; library users import the same names from `agentic-mermaid/agent` — including `describeOps(family)` and `opSignatures(family)`, which return each op's exact fields, required-ness, inlined enum values, and constraint/default notes (e.g. journey score `integer 1..5`, flowchart shape `default: rectangle`) so you can look up an unfamiliar op instead of guessing; use `describeMermaidFacts` / `checkMermaid` when you need semantic read-back such as `edge Processing -> [*] : done`, `member Duck +quack()`, or `task Docs start after core`. A hosted MCP at `agentic-mermaid.dev/mcp` (stateless Streamable HTTP JSON-RPC) exposes `execute` (Code Mode), `render_svg`, `render_ascii`, `render_png`, `verify`, `describe`, and the declarative `mutate` (edit a `source`) and `build` (author from a `family`) — which apply a JSON op list and return `{ok,family,source,verify}`, prescriptive `{ok:false,…,error}` on a bad op; prefer them for structured edits and reserve `execute` for logic the ops do not express — all with 64KB input caps; the call shape is a plain POST (stateless, no initialize handshake): `POST /mcp` with `content-type: application/json` and body `{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"verify","arguments":{"source":"flowchart TD\n  A --> B"}}}`. Prefer the local library, CLI, or a self-hosted MCP and reach for the hosted endpoint only when you cannot install. Agentic Mermaid outputs ASCII, PNG, and SVG; Unicode text and JSON layout are also available. Flowchart authoring facts: quote any label carrying punctuation (`id["HTTPS /api/sessions*"]`); `\n` inside a quoted label is a line break and canonicalizes to `<br>` on serialize; `subgraph id["Title"] … end` groups nodes; labeled and dotted edges are `A -- "label" --> B`, `A -.-> B`, and `A -. "label" .-> B`. For every other family, read that family's exact header, edge, and relationship syntax from its `example` in `am capabilities --json` (`capabilities.json` `families[].example`) rather than assuming flowchart syntax — e.g. class inheritance `Base <|-- Derived`, ER cardinality `A ||--o{ B : label`, architecture edges `api:R --> L:db`.
@@ -51,7 +57,7 @@ Tier 1 (structural, reliable, universal): `EMPTY_DIAGRAM`, `EDGE_MISANCHORED`, `
 
 Tier 2 (geometric, advisory, flowchart-specific): `NODE_OVERLAP`, `ROUTE_SELF_CROSS`, plus the route-contract tripwires `ROUTE_HITCH`, `ROUTE_UNEXPLAINED_BEND`, `ROUTE_LABEL_ON_SHARED_TRUNK`, `ROUTE_CONTAINER_MISANCHOR`, `ROUTE_SHAPE_MISANCHOR`, `ROUTE_STALE_AFTER_NODE_MOVE` (docs/design/system/route-contracts.md — the layout pipeline upholds these itself, so they fire only on pipeline regressions). Route tripwires fire for flowchart/state; class and ER additionally run boundary-anchor and overlap checks (`ROUTE_SHAPE_MISANCHOR`, `NODE_OVERLAP` on class/entity boxes). For other families, geometric concerns surface via perceptual metrics (`measureQuality(layoutMermaid(d))`). See `docs/quality.md`. Don't gate CI on Tier 2 alone.
 
-Tier 3 (lint, advisory): `DUPLICATE_EDGE`, `UNREACHABLE_NODE`, `DECISION_BRANCH_UNLABELED`, `COMMENT_DROPPED`, `UNSUPPORTED_SYNTAX`, `CONTENT_DROPPED_ON_ROUNDTRIP`. These catch common agent mistakes, source-preserved Mermaid syntax that is not fully modeled, and content-bearing sources whose local layout renders nothing. They never flip `verify.ok`; inspect and fix when the task asks for clean maintainable diagrams. `DECISION_BRANCH_UNLABELED` means a multi-exit decision diamond has an unlabeled branch. `COMMENT_DROPPED` means in-body `%%` comments will not survive structured serialization; the leading wrapper (frontmatter, `%%{init}%%` directives, comments before the header) always round-trips byte-verbatim. `UNSUPPORTED_SYNTAX` means either a Mermaid construct (for example flowchart edge IDs, edge metadata, click/href, or markdown strings) is preserved losslessly as source but is not fully modeled by local structured mutation/render semantics, or `syntax: "empty_layout"` means the source carries content but the local layout produced a 0x0 canvas with no nodes, edges, or groups. Inspect the warning message and `verify.layout` before accepting the artifact. `CONTENT_DROPPED_ON_ROUNDTRIP` means the structured node/edge/group tally changed across a parse → serialize → re-parse cycle — canonical serialization is silently dropping or duplicating content even though the bytes may re-parse; treat it as a faithfulness bug to report.
+Tier 3 (lint, advisory): `DUPLICATE_EDGE`, `UNREACHABLE_NODE`, `DECISION_BRANCH_UNLABELED`, `COMMENT_DROPPED`, `UNSUPPORTED_SYNTAX`, `CONTENT_DROPPED_ON_ROUNDTRIP`, `INEFFECTIVE_CONFIG`. These catch common agent mistakes, source-preserved Mermaid syntax that is not fully modeled, and content-bearing sources whose local layout renders nothing. They never flip `verify.ok`; inspect and fix when the task asks for clean maintainable diagrams. `DECISION_BRANCH_UNLABELED` means a multi-exit decision diamond has an unlabeled branch. `COMMENT_DROPPED` means in-body `%%` comments will not survive structured serialization; the leading wrapper (frontmatter, `%%{init}%%` directives, comments before the header) always round-trips byte-verbatim. `UNSUPPORTED_SYNTAX` means either a Mermaid construct (for example flowchart edge IDs, edge metadata, click/href, or markdown strings) is preserved losslessly as source but is not fully modeled by local structured mutation/render semantics, or `syntax: "empty_layout"` means the source carries content but the local layout produced a 0x0 canvas with no nodes, edges, or groups. Inspect the warning message and `verify.layout` before accepting the artifact. `CONTENT_DROPPED_ON_ROUNDTRIP` means the structured node/edge/group tally changed across a parse → serialize → re-parse cycle — canonical serialization is silently dropping or duplicating content even though the bytes may re-parse; treat it as a faithfulness bug to report. `INEFFECTIVE_CONFIG` means a Mermaid config field was accepted for compatibility but has no effect on this family's output (the warning names the field).
 
 ## CLI verbs
 

package/README.md

@@ -136,7 +136,7 @@ For multi-step MCP edits, connect `agentic-mermaid-mcp` and use Code Mode `execu
 
 Agentic Mermaid ships a Model Context Protocol server so MCP-capable agents can render and safely edit diagrams without shelling out.
 
-- **Self-hosted (default).** `agentic-mermaid-mcp` runs a stdio server exposing `execute` (Code Mode sandbox), `render_png`, and `describe`. Add `--transport http` for HTTP/SSE with managed PNG file/URL artifacts. See [`docs/mcp-http-transport.md`](./docs/mcp-http-transport.md) and [`docs/mcp-code-mode-rationale.md`](./docs/mcp-code-mode-rationale.md).
+- **Self-hosted (default).** `agentic-mermaid-mcp` runs a stdio server exposing `execute` (Code Mode sandbox), `render_png`, and `describe`. Package runners can use `npx -y agentic-mermaid mcp`; the `mcp` argument routes the package-name binary to the same stdio server. Add `--transport http` for HTTP/SSE with managed PNG file/URL artifacts. See [`docs/mcp-http-transport.md`](./docs/mcp-http-transport.md) and [`docs/mcp-code-mode-rationale.md`](./docs/mcp-code-mode-rationale.md).
 - **Hosted.** A stateless Streamable HTTP endpoint is available at `https://agentic-mermaid.dev/mcp` (tools: `execute`, `render_svg`, `render_ascii`, `render_png`, `verify`, `describe`, `mutate`, `build`; 64 KB input caps). It is MCP JSON-RPC only, not a REST render API. Hosted `execute` runs the same `mermaid.*` facade in a Cloudflare Dynamic Worker isolate with no network; hosted PNG returns base64 only.
 
 Local-first is the default posture: prefer the library, CLI, or a self-hosted MCP for anything sensitive, offline, larger than the hosted caps, or needing local PNG file/URL artifacts. The hosted endpoint is a public, unauthenticated convenience for zero-install render/verify/describe and bounded structured edits.

package/TODO.md

@@ -39,23 +39,23 @@ dependents after. IDs are stable names, not an ordering.
   `{ "ok": true, "value": 42, "logs": [] }`, proving the production Dynamic
   Workers `LOADER` binding resolves for hosted Code Mode. The pure tools are no
   longer the only production evidence.
-- [ ] **DEC-4 — Submit the sitemap to Google and Bing** (`owner-decision`).
-  After the first deploy that serves `https://agentic-mermaid.dev/sitemap.xml`
-  (generated by `website/build.ts`, validated against the sitemaps.org 0.9
-  XSD), submit it manually in Google Search Console → Sitemaps and Bing
-  Webmaster Tools → Sitemaps. Both require a signed-in, ownership-verified
-  account, and the anonymous ping endpoints are retired, so this cannot be
-  automated in the repo.
-- [ ] **DEC-5 — Add the `Sitemap:` line in the Cloudflare dashboard**
-  (`owner-decision`). Production serves Cloudflare's *managed* content-signals
-  `robots.txt` (verified live: the "content signals" template, no `Sitemap:`
-  line), which the edge is likely to serve in preference to any repo
-  `website/public/robots.txt`. So the reliable way to make the sitemap
-  auto-discoverable is to add `Sitemap: https://agentic-mermaid.dev/sitemap.xml`
-  via the Cloudflare dashboard's managed robots.txt (Manage robots.txt), which
-  also keeps the content-signals reservation intact. The repo deliberately ships
-  no `robots.txt` (it would only collide with the managed one), so the dashboard
-  is the single source for this line.
+- [ ] **DEC-4 — Establish Google and Bing search visibility**
+  (`owner-decision`). In ownership-verified Google Search Console and Bing
+  Webmaster Tools, submit `https://agentic-mermaid.dev/sitemap.xml`, request
+  indexing for the homepage and core docs, and monitor coverage plus exact-name
+  queries for "Agentic Mermaid" and "agentic-mermaid". Fix crawl/canonical
+  findings before pursuing relevant external references and backlinks; do not
+  manufacture low-quality links. These consoles require a signed-in owner and
+  the anonymous sitemap ping endpoints are retired, so this cannot be automated
+  in the repo.
+- [ ] **DEC-5 — Finish Cloudflare's managed `robots.txt` policy**
+  (`owner-decision`). Production serves Cloudflare's managed content-signals
+  file in preference to a repository asset. In **Manage robots.txt**, choose the
+  intended crawler/content-signals policy, preserve access for the public pages
+  meant to be indexed, and add
+  `Sitemap: https://agentic-mermaid.dev/sitemap.xml`. Then verify the live body
+  and status with `curl` and both search consoles. The repo deliberately ships
+  no competing `robots.txt`, so the dashboard remains the single source.
 
 ## 1. Security backlog
 
@@ -78,18 +78,31 @@ dependents after. IDs are stable names, not an ordering.
 - [ ] **BUILD-27 — MCP Apps support** (`todo`). Expose an interactive
   in-agent diagram UI through MCP Apps: `ui://` resources, correct
   `text/html;profile=mcp-app` resource MIME type, tool `_meta.ui.resourceUri`
-  wiring, CSP for the view, and tests that the resources are reachable without
-  leaking secrets. Start with a read-only preview/verify view before adding any
-  editable surface.
-- [ ] **BUILD-28 — WebMCP support** (`todo`). Track and implement the WebMCP
-  browser-discovery path once the draft stabilizes enough to commit to a public
-  contract. Scope should cover `/.well-known/mcp` compatibility, browser-safe
-  CORS/Origin handling, and parity with the hosted Streamable HTTP tool surface.
-- [ ] **BUILD-29 — ChatGPT app listing** (`todo`). Prepare a ChatGPT app
-  listing once Agentic Mermaid has a stable in-agent UI surface. Package the
-  public description, icon, privacy/security notes, hosted MCP/App endpoints,
-  and reviewer-ready examples; do not submit until the MCP Apps/WebMCP surface
-  is production-ready.
+  wiring, resource CSP/domain metadata, and tests that the resources are
+  reachable without leaking secrets. Start with a portable, read-only
+  preview/verify view before adding an editable surface; keep tool results useful
+  in MCP hosts that ignore the UI extension.
+- [ ] **BUILD-28 — Experimental page-local WebMCP support** (`todo`). The
+  current Web Machine Learning Community Group report is not a W3C Standard or
+  Standards Track document. In the browser editor, feature-detect
+  `document.modelContext` and register a narrow active-document surface with
+  `document.modelContext.registerTool()` using exact JSON Schemas, truthful
+  `readOnlyHint` / `untrustedContentHint` annotations, and an `AbortSignal` for
+  lifecycle cleanup. Start with read source, verify, describe, and render; add
+  structured mutation only after explicit user-action and state-consistency
+  tests. Gate it behind supported-browser detection and test with both a shim
+  and a compatible browser trial. This is not `/.well-known/mcp`, CORS, or
+  Streamable HTTP parity: the draft is an in-page browser API and does not
+  prescribe MCP as the browser agent's transport.
+- [ ] **BUILD-29 — Submit a ChatGPT app as a plugin** (`todo`, after
+  BUILD-27; WebMCP is not a dependency). Validate the public MCP + MCP Apps
+  experience in ChatGPT Developer Mode on web and mobile, define an exact CSP,
+  and audit tool schemas, `_meta`, instructions, and annotations before scanning
+  the endpoint. Complete organization verification and `api.apps.write`
+  access, then prepare the plugin submission's name, logo, description, company
+  and privacy-policy URLs, screenshots, test prompts/responses, localization,
+  and review notes. Submit through the plugin portal only when the live endpoint
+  and UI are stable enough to preserve the reviewed metadata contract.
 - [ ] **BUILD-25 — Structured coverage for the last silent-opaque constructs**
   (`todo`, needs legacy-renderer support first; tracked as **#118**). The systemic `UNSUPPORTED_SYNTAX`
   warning (2026-07) now announces every opaque fallback, so these are visible but
@@ -560,8 +573,17 @@ _No active blocked items._
 
 ## 5. Parked / evidence-required ideas
 
-- [ ] **PARK-2 — `.well-known/skills` discovery** (`parked`). Watch the
-  ecosystem; do not implement until a standard settles.
+- [ ] **PARK-2 — Agent Skills discovery and skills.sh visibility** (`parked`,
+  experimental draft). Reassess Cloudflare's Agent Skills Discovery draft when
+  its `agent-skills` well-known suffix is registered or the ecosystem contract
+  stabilizes. If adopted earlier, generate a v0.2.0
+  `/.well-known/agent-skills/index.json`, package the workflow skill plus its
+  references as a safe archive, publish its SHA-256 digest, support GET/HEAD +
+  JSON/Markdown/archive content types + CORS/cache headers, and test fetch,
+  extraction, and digest verification against the published schema. Once that
+  endpoint exists, change the homepage `Link` header from the direct `SKILL.md`
+  to the index. Separately document `npx skills add adewale/agentic-mermaid`;
+  skills.sh has no submission API and gains visibility from real CLI installs.
 - [ ] **PARK-3 — Fork feature ports** (`parked`). Vercel themes,
   browser/package export tweaks, C4, ArchiMate (upstream PR #34), and
   animation remain fork-audit ideas. Promote one only with a focused issue

package/assets/fonts/FONT-LICENSES.md

@@ -2,11 +2,14 @@
 
 The `.ttf` files in this directory ship inside the `agentic-mermaid` npm
 package so PNG rasterization (`renderMermaidPNG`) is deterministic and
-offline: DejaVu Sans is the default face, and the four OFL faces are the
-typefaces referenced by the built-in styles (hand-drawn, excalidraw,
-watercolor → Caveat; pen-and-ink, tufte → EB Garamond; freehand →
-Architects Daughter; blueprint → Share Tech Mono). Custom styles that
-reference other families can supply them via the `fontDirs` PNG option.
+offline: Inter (weights 400/500/600/700 — the weights the SVG output
+requests) is the default face and the family the text-metrics model is
+calibrated for; DejaVu Sans serves as the per-glyph fallback for symbols
+Inter lacks; and the remaining OFL faces are the typefaces referenced by
+the built-in styles (hand-drawn, excalidraw, watercolor → Caveat;
+pen-and-ink, tufte → EB Garamond; freehand → Architects Daughter;
+blueprint → Share Tech Mono). Custom styles that reference other families
+can supply them via the `fontDirs` PNG option.
 
 ## OFL-1.1 faces
 
@@ -16,6 +19,7 @@ required by OFL condition 2.
 
 | File | Typeface | Copyright / Reserved Font Name | Source |
 |---|---|---|---|
+| `Inter-Regular.ttf`, `Inter-Medium.ttf`, `Inter-SemiBold.ttf`, `Inter-Bold.ttf` | Inter (v4.001 statics, weights 400/500/600/700) | Copyright 2016 The Inter Project Authors (https://github.com/rsms/inter) | fonts.google.com/specimen/Inter |
 | `Caveat.ttf` | Caveat | Copyright 2014 The Caveat Project Authors (https://github.com/googlefonts/caveat) | fonts.google.com/specimen/Caveat |
 | `EBGaramond.ttf` | EB Garamond | Copyright 2017 The EB Garamond Project Authors (https://github.com/octaviopardo/EBGaramond12) | fonts.google.com/specimen/EB+Garamond |
 | `ArchitectsDaughter.ttf` | Architects Daughter | Copyright (c) 2010, Kimberly Geswein (kimberlygeswein.com) | fonts.google.com/specimen/Architects+Daughter |

package/dist/agent.d.ts

@@ -1,5 +1,5 @@
-import { N as NodeShape, E as EdgeStyle, M as MermaidFrontmatterMap, a as MermaidConfigMap, b as MermaidGraph, D as Direction, R as RouteClass, c as EdgeRouteCertificate, d as RegionContainmentCertificate, e as MermaidNode, f as MermaidEdge, g as NormalizedMermaidSource, h as RenderOptions, i as DiagramColors, P as PositionedDiagram, j as RenderContext, S as SceneDoc, A as AsciiConfig, C as ColorMode, k as AsciiTheme, L as LayoutRouteCertificate, l as AsciiRenderOptions, r as renderMermaidASCII$1, m as renderMermaidSVG$1, n as StyleInput } from './index-x2Cj1dVv.js';
-export { o as AnyPort, p as DiamondFacet, F as FamilyEdgeRouteCertificate, q as FamilyRouteCertificate, s as LayoutRouteClass, t as PortSemanticRole, u as PortSide, v as RouteBlocker, w as RouteCertificate, x as RoutePortAssignment, y as StyleSpec, T as TEXT_MEASUREMENT_CONTRACT, z as TextMeasurementContract, B as TextMeasurementInput, G as TextMeasurementResult, H as getStyle, I as inferBackend, J as knownStyles, K as measureText, O as measureTextWidth, Q as registerStyle, U as resolveStyleStack, V as validateStyleSpec, W as verifyNoExternalRefs } from './index-x2Cj1dVv.js';
+import { N as NodeShape, E as EdgeStyle, M as MermaidFrontmatterMap, a as MermaidConfigMap, b as MermaidGraph, D as Direction, R as RouteClass, c as EdgeRouteCertificate, d as RegionContainmentCertificate, e as MermaidNode, f as MermaidEdge, g as NormalizedMermaidSource, h as RenderOptions, i as DiagramColors, P as PositionedDiagram, j as RenderContext, S as SceneDoc, A as AsciiConfig, C as ColorMode, k as AsciiTheme, L as LayoutRouteCertificate, l as AsciiRenderOptions, r as renderMermaidASCII$1, m as renderMermaidSVG$1, n as StyleInput } from './index-wjRjbsP-.js';
+export { o as AnyPort, p as DiamondFacet, F as FamilyEdgeRouteCertificate, q as FamilyRouteCertificate, s as LayoutRouteClass, t as PortSemanticRole, u as PortSide, v as RouteBlocker, w as RouteCertificate, x as RoutePortAssignment, y as StyleSpec, T as TEXT_MEASUREMENT_CONTRACT, z as TextMeasurementContract, B as TextMeasurementInput, G as TextMeasurementResult, H as getStyle, I as inferBackend, J as knownStyles, K as measureText, O as measureTextWidth, Q as registerStyle, U as resolveStyleStack, V as validateStyleSpec, W as verifyNoExternalRefs } from './index-wjRjbsP-.js';
 
 type Result<T, E> = {
     ok: true;
@@ -691,6 +691,7 @@ type JourneyMutationOp = {
 } | {
     kind: 'add_section';
     label: string;
+    index?: number;
 } | {
     kind: 'remove_section';
     index: number;
@@ -704,6 +705,7 @@ type JourneyMutationOp = {
     text: string;
     score: number;
     actors?: string[];
+    index?: number;
 } | {
     kind: 'remove_task';
     sectionIndex: number;
@@ -727,6 +729,22 @@ type JourneyMutationOp = {
     kind: 'rename_actor';
     from: string;
     to: string;
+} | {
+    kind: 'move_task';
+    fromSection: number;
+    fromIndex: number;
+    toSection: number;
+    toIndex: number;
+} | {
+    kind: 'move_section';
+    from: number;
+    to: number;
+} | {
+    kind: 'set_accessibility_title';
+    title: string | null;
+} | {
+    kind: 'set_accessibility_description';
+    description: string | null;
 };
 type ArchitectureMutationOp = {
     kind: 'add_service';
@@ -968,7 +986,7 @@ type Tier2WarningCode = 'NODE_OVERLAP' | 'ROUTE_SELF_CROSS' | 'ROUTE_HITCH' | 'R
  * Tier 3 (advisory lint). Family-specific quality hints for common agent
  * mistakes that still parse and render. Lint warnings never flip verify.ok.
  */
-type Tier3WarningCode = 'DUPLICATE_EDGE' | 'UNREACHABLE_NODE' | 'DECISION_BRANCH_UNLABELED' | 'COMMENT_DROPPED' | 'UNSUPPORTED_SYNTAX' | 'CONTENT_DROPPED_ON_ROUNDTRIP';
+type Tier3WarningCode = 'DUPLICATE_EDGE' | 'UNREACHABLE_NODE' | 'DECISION_BRANCH_UNLABELED' | 'COMMENT_DROPPED' | 'UNSUPPORTED_SYNTAX' | 'CONTENT_DROPPED_ON_ROUNDTRIP' | 'INEFFECTIVE_CONFIG';
 type WarningCode = Tier1WarningCode | Tier2WarningCode | Tier3WarningCode;
 type LayoutWarning = {
     code: 'EMPTY_DIAGRAM';
@@ -1093,6 +1111,17 @@ type LayoutWarning = {
         edges: number;
         groups: number;
     };
+}
+/**
+ * A Mermaid config field was accepted (for config-shape compatibility) but
+ * has no effect on this family's geometry or paint — e.g. Journey's
+ * sequence-era fields. Accepting-and-ignoring silently misleads migrating
+ * users; this lint says so. Never flips verify.ok.
+ */
+ | {
+    code: 'INEFFECTIVE_CONFIG';
+    field: string;
+    message: string;
 };
 declare const WARNING_SEVERITY: Record<WarningCode, WarningSeverity>;
 declare const WARNING_TIER: Record<WarningCode, WarningTier>;
@@ -1540,7 +1569,7 @@ declare const MUTATION_OPS_BY_FAMILY: {
     readonly timeline: readonly ["set_title", "add_section", "remove_section", "set_section_label", "add_period", "remove_period", "set_period_label", "add_event", "remove_event", "set_event_text"];
     readonly class: readonly ["set_title", "add_class", "remove_class", "rename_class", "add_member", "remove_member", "add_relation", "remove_relation", "add_note", "remove_note"];
     readonly er: readonly ["add_entity", "remove_entity", "rename_entity", "add_attribute", "remove_attribute", "add_relation", "remove_relation"];
-    readonly journey: readonly ["set_title", "add_section", "remove_section", "set_section_label", "add_task", "remove_task", "set_task_text", "set_task_score", "set_task_actors", "rename_actor"];
+    readonly journey: readonly ["set_title", "add_section", "remove_section", "set_section_label", "add_task", "remove_task", "set_task_text", "set_task_score", "set_task_actors", "rename_actor", "move_task", "move_section", "set_accessibility_title", "set_accessibility_description"];
     readonly architecture: readonly ["add_service", "remove_service", "rename_service", "set_service_label", "set_service_icon", "move_service", "add_group", "remove_group", "add_edge", "remove_edge"];
     readonly xychart: readonly ["set_title", "set_x_axis", "set_y_axis", "add_series", "remove_series", "set_series_values", "set_series_name", "reorder_series"];
     readonly pie: readonly ["set_title", "set_show_data", "add_slice", "remove_slice", "rename_slice", "set_slice_value", "reorder_slice"];
@@ -1801,6 +1830,21 @@ interface LayoutMermaidOptions {
 }
 declare function layoutMermaid(d: ValidDiagram, opts?: LayoutMermaidOptions): RenderedLayout;
 
+/**
+ * Loud-tofu guard: emitted when rendered text contains characters no loaded
+ * font covers (they would silently rasterize as empty boxes). Delivered to
+ * `PngOptions.onWarning` when provided, otherwise written to stderr — never
+ * silent either way.
+ */
+interface PngFontWarning {
+    code: 'PNG_FONT_COVERAGE';
+    /** Coarse script bucket of the uncovered characters ('CJK', 'emoji', …). */
+    script: string;
+    /** The uncovered characters, unique, ascending codepoint order. */
+    chars: string[];
+    /** Human-readable summary naming the script and the escape hatches. */
+    message: string;
+}
 interface PngOptions {
     /** Output scale multiplier (default 2 — retina). */
     scale?: number;
@@ -1818,8 +1862,18 @@ interface PngOptions {
     /** Ink-wobble seed for styled looks, same as RenderOptions.seed. */
     seed?: number;
     /** Extra font directories searched in addition to the bundled ones —
-     *  the escape hatch for custom styles that reference unbundled families. */
+     *  the escape hatch for custom styles that reference unbundled families
+     *  and for scripts the bundled fonts don't cover (CJK, emoji). */
     fontDirs?: string[];
+    /** Also load the OS's installed fonts (default false). Opting in trades
+     *  cross-machine determinism for coverage — useful when the diagram needs
+     *  CJK/emoji glyphs and a system font provides them. Coverage warnings are
+     *  skipped in this mode: the system font set is unknown to the checker. */
+    loadSystemFonts?: boolean;
+    /** Receives glyph-coverage warnings instead of the default stderr write.
+     *  Warnings never change the PNG bytes; identical inputs render
+     *  identically whether or not a handler is installed. */
+    onWarning?: (warning: PngFontWarning) => void;
 }
 /**
  * Render a Mermaid diagram (source string or ValidDiagram) to a PNG byte array.