agentleFS
lintendolintendo/Axhub-Skills/skills/axhub-prototype-context/SKILL.md

axhub-prototype-context

Skill145 starsChanged 19 days agoMITView on GitHub

Use when an annotated prototype URL exposes PRD, directory, annotation, or source context that an AI Agent needs for review, implementation planning, production work, source handoff, or optional review-report submission.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Wait until the app renders, then poll briefly for window.__AXHUB_ANNOTATION_SOURCE__.

    line 13
  2. Very specific:

    When full Markdown annotation text is needed, also read window.__AXHUB_ANNOTATION_SOURCE_DOCUMENT__?.markdownMap.

    line 15
  3. Distinctive:

    If the snapshot is missing, report the URL, page title, relevant console errors, and that the annotated prototype context snapshot was not …

    line 16
  4. Very specific:

    If the user asks to submit a review report back to Axhub Make, read references/review-report-submission.md.

    line 17

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Annotated Prototype Context

Read the published context from an annotated prototype so an AI Agent can understand its pages, PRD, annotations, and available source before continuing with review or production work. Treat the page as read-only unless the user asks to submit a review report back to Axhub Make.

Workflow

  1. Open the requested annotated prototype URL with Playwright, Browser, Chrome, or an equivalent tool that can evaluate page JavaScript.
  2. Wait until the app renders, then poll briefly for window.__AXHUB_ANNOTATION_SOURCE__.
  3. Evaluate and return that value. Do not modify the object or write anything back to window.
  4. When full Markdown annotation text is needed, also read window.__AXHUB_ANNOTATION_SOURCE_DOCUMENT__?.markdownMap; report when it is unavailable.
  5. If the snapshot is missing, report the URL, page title, relevant console errors, and that the annotated prototype context snapshot was not published.
  6. If the user asks to submit a review report back to Axhub Make, read references/review-report-submission.md.

Minimal page evaluation:

await page.waitForFunction(() => window.__AXHUB_ANNOTATION_SOURCE__, { timeout: 10000 });
const source = await page.evaluate(() => window.__AXHUB_ANNOTATION_SOURCE__);

What To Report

  • Page navigation from source.directory (folder, route, and link nodes).
  • PRD/document outline and article bodies from source.documents (folder, markdown, and html nodes). Read each Markdown article's inline markdown and each HTML article's inline html; report a missing body when only markdownPath or htmlPath is present. Also check legacy directory document nodes when present.
  • Annotation count and the important node fields: id, title, pageId, locator, annotationText, aiPrompt, color, and controls.
  • Markdown annotation bodies for source.nodes with hasMarkdown: true are in the source document's markdownMap[node.id]; do not mistake them for PRD articles. Include source.targets when explaining document links to unannotated page elements.
  • Source handoff from source.source when present. Report root and manifest; when source is needed, read the manifest and relevant files via root.
  • Mention whether images are attached by checking images.length; do not download images unless the user asks.

Data Shape

type AnnotationSourceRuntimeSnapshot = {
  directory: AnnotationDirectory | null;
  documents: AnnotationDocumentDirectory | null;
  nodes: AnnotationNode[];
  targets?: AnnotationElementTarget[];
  presentation?: AnnotationSourcePresentation;
  source?: {
    root?: string;
    manifest?: string;
  };
};
  • directory.nodes is the page navigation tree. New PRD documents belong in documents.nodes; older sources may still contain markdown nodes in directory.
  • documents.nodes contains folder, markdown, and html nodes. Readable Markdown nodes have an inline markdown body and may include markdownPath; HTML nodes have an inline html body and may include htmlPath. In Make, these files typically live under src/prototypes/<prototype-id>/docs/. The host must inline their contents before publishing the snapshot; the runtime does not fetch document files from their paths.
  • readerMode: "full" | "split" controls an article's initial layout; omitted values default to split. Documents and pages have separate directory entries.
  • Relative links between articles resolve through unique, normalized markdownPath or htmlPath values. Standard HTML #anchor links are bridged to the reader so they use the same controlled in-document navigation as Markdown, while a link such as [View element](#axhub-target:element-id) refers to an annotation node or a lightweight targets entry in both Markdown and HTML. Cross-page positioning requires host routing (onTargetRoute, or a matching directory route fallback); the snapshot alone does not establish that navigation works.
  • nodes is the full annotation list, independent of current page, selected state, and color filter.
  • Each annotation node includes id, index, optional title, optional pageId, locator, aiPrompt, annotationText, hasMarkdown, color, images, optional controls, createdAt, and updatedAt.
  • source is only a source-code discovery reference. In HTML published with source included, manifest points to source/manifest.json; resolve its paths relative to source.root.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.