agentleFS
samulee003samulee003/agentic-ppt-skills/.agents/skills/current-slide/SKILL.md

current-slide

Skill0 starsChanged 25 days agoMITView on GitHub

Resolve which slide, page, and (optionally) selected element the user is currently viewing in the open-slide dev server. Consult this whenever the user references "this page", "this slide", "this element", "the slide I'm on", "the current page", or any deictic reference to slide content without naming it. Re-read `node_modules/.open-slide/current.json` at the start of every such turn — the user navigates between turns, so a value you read earlier in the conversation is almost certainly stale.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 3 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before asking "which slide?" or "which element?" as a clarifying question — check this file first.

    line 65
  2. Distinctive:

    Before guessing from git log, recently-edited files, or the most recent slide folder.

    line 66
  3. Distinctive:

    Don't create the file or guess.

    line 87
  4. Very specific:

    Check updatedAt is recent.

    line 94
  5. Very specific:

    Read pagePath (e.g. slides/q2-roadmap/index.tsx).

    line 95
  6. Very specific:

    If current.json is missing or stale, ask: "Which slide and page should I tighten?

    line 99
  7. Very specific:

    If selection is non-null, the user means that element.

    line 106
  8. Very specific:

    If selection is null, fall back to the page-level flow above — and consider asking "which element?" since the user used a deictic but …

    line 110

≈ “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

Where is the user right now?

When the user says "fix this page", "tweak this heading", or "the slide I'm looking at", they almost never name the slide id, page number, or element — they mean wherever they are in the dev viewer. Before asking "which slide?" or "which element?", check the file the dev server writes on every navigation and inspector pick.

Re-read on every deictic turn — never reuse a prior read

current.json is a live cursor, not a fact about the conversation. The user moves between slides, pages, and elements freely between your turns — including while you were doing other work. Read the file fresh at the start of every new turn that uses a deictic reference, even if:

  • you already read it earlier in this same conversation,
  • you just finished editing the slide it pointed to,
  • the user's new message sounds like a continuation ("now make it bigger", "also fix this one", "keep going").

A "continue editing" follow-up is exactly the case where the user has likely just navigated to a different slide or picked a different element. Trusting your last read here will silently edit the wrong file. Re-read, compare slideId / pageIndex / selection against what you used last time, and act on the new values.

How to read it

node_modules/.open-slide/current.json

Path is relative to the project root (the user's cwd, the directory that contains slides/ and package.json). Use the Read tool. The file is JSON.

What you get

{
  "slideId": "q2-roadmap",
  "pageIndex": 2,
  "pageNumber": 3,
  "totalPages": 8,
  "slideTitle": "Q2 Roadmap",
  "view": "slides",
  "pagePath": "slides/q2-roadmap/index.tsx",
  "selection": {
    "line": 42,
    "column": 6,
    "tagName": "h1",
    "text": "Q2 Roadmap"
  },
  "updatedAt": "2026-05-09T14:32:11.123Z"
}
  • slideId — folder name under slides/. Use as-is for any /__slides/<id>/... API or as the URL segment.
  • pageIndex — 0-based, for use with the page array in index.tsx (export default [Cover, Body, ...]).
  • pageNumber — 1-based, for use in messages to the user ("page 3 of 8") and for the URL ?p=N.
  • pagePath — relative path to the slide source. Hand straight to Read / Edit.
  • view — "slides" (canvas view) or "assets" (asset manager). If "assets", the user is browsing files for that slide rather than viewing a page.
  • selection — null if nothing is selected. Otherwise, the JSX element the user picked in the inspector overlay:
    • line (1-indexed) and column (0-indexed) point to the JSX opening tag inside pagePath. This is the canonical handle — match against the source line, not the rendered DOM.
    • tagName is the rendered DOM tag, lowercased ("h1", "div", "img").
    • text is a trimmed text snippet (≤120 chars) from the element's textContent, useful as a sanity check that you're looking at the right node.
    • Selection auto-clears whenever the user navigates to a different slide or page.
  • updatedAt — ISO timestamp of the last navigation or selection change. Use it to detect staleness.

When to use this

  • The user references the current slide/page deictically: "this", "here", "the page I'm on", "the slide I'm looking at", "what I'm working on".
  • The user references a specific element: "this heading", "this image", "the button I just clicked", "tighten this", "change the color of this". If selection is non-null, that's the element they mean.
  • Before asking "which slide?" or "which element?" as a clarifying question — check this file first.
  • Before guessing from git log, recently-edited files, or the most recent slide folder.

When NOT to use this

  • The user names a slide explicitly ("edit q2-roadmap") — use that name directly.
  • The apply-comments workflow already finds the right file via @slide-comment markers; it doesn't need this skill.
  • For listing or discovering slides — read slides/ directly.

Staleness — verify before acting

updatedAt is the last time the user navigated. Treat it like a cache:

  • Fresh (under ~5 minutes old): trust it. Open pagePath, do the work.
  • Older than ~5 minutes: confirm with the user before editing. The dev server may not be running; the user may have switched contexts.
  • Hours/days old: ignore it. Ask the user which slide they mean.

A newer updatedAt than the one you saw last turn is the normal signal that the user has moved — switch to the new slideId / pageIndex / selection without asking.

When the file is missing

  • The dev server hasn't been opened on a slide yet, or has never run.
  • Don't create the file or guess. Ask the user which slide they mean, or suggest they open the slide in the dev server first.

Example — page-level reference

User: "tighten the spacing on this page"

  1. Read node_modules/.open-slide/current.json.
  2. Check updatedAt is recent.
  3. Read pagePath (e.g. slides/q2-roadmap/index.tsx).
  4. Identify the page at pageIndex in the default-exported array.
  5. Consult the slide-authoring skill for spacing rules, then edit that page in place.

If current.json is missing or stale, ask: "Which slide and page should I tighten? The dev server hasn't published a current page recently."

Example — element-level reference

User: "make this bigger"

  1. Read node_modules/.open-slide/current.json.
  2. If selection is non-null, the user means that element. Read pagePath, jump to selection.line, and find the JSX opening tag near that line/column. Confirm with the snippet in selection.text and the tagName.
  3. Consult slide-authoring for type-scale and layout rules before editing.
  4. Edit the JSX node in place.

If selection is null, fall back to the page-level flow above — and consider asking "which element?" since the user used a deictic but hasn't picked one in the inspector.

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.