Browser UseBrowser Skills
← Back to skills

Domain skill

figma

Markdown synced from browser-harness domain skills.

Host
figma
Files
1

Agent prompt

Use this skill

Copy this prompt into your coding agent to make it enable browser-harness domain skills and read this exact domain folder before automating.

Set up https://github.com/browser-use/browser-harness for me if it is not already installed. If setup is needed, read `install.md` first to install and connect it to my real browser. Then read `SKILL.md` for normal usage and always read `helpers.py` because that is where the browser-harness functions are.

Enable domain skills if they are not already enabled by setting `BH_DOMAIN_SKILLS=1` for browser-harness. Use the `figma` domain skill from `agent-workspace/domain-skills/figma/`. Read every markdown file for this domain before inventing an approach:
- agent-workspace/domain-skills/figma/comments.md

Use those domain-skill notes to complete my task for `figma` in my real browser. When you open a setup, verification, or task tab, activate it so I can see the active browser tab.

Skill contents

What the agent will read

reading file comments

comments.md

Source
  • Figma comments are invisible to the Plugin API — plugins cannot read them at all. The official REST endpoint (api.figma.com/v1/files/:key/comments) needs a personal access token and is metered. But the web app's own...
  • Same-origin from any www.figma.com page, so call it with js() from wherever you already are:
  • The file key is the middle segment of any share URL: figma.com/design/<FILEKEY>/<slug>.
  • { meta: [...] } — one flat array, every comment in the file (threads included). Useful fields per comment:
Show full markdown

Figma comments are invisible to the Plugin API — plugins cannot read them at all. The official REST endpoint (api.figma.com/v1/files/:key/comments) needs a personal access token and is metered. But the web app's own internal API serves the full comment set with nothing but the session cookies already in the browser.

The endpoint

code
GET https://www.figma.com/api/file/<FILE_KEY>/comments

Same-origin from any www.figma.com page, so call it with js() from wherever you already are:

python
r = js("""
(async () => {
  const response = await fetch("/api/file/<FILE_KEY>/comments");
  if (!response.ok) throw new Error("Comments request failed: " + response.status);
  const data = await response.json();
  if (!Array.isArray(data.meta)) throw new Error("Comments response has no list");
  const byId = Object.fromEntries(data.meta.map(c => [c.id, c]));
  return data.meta.filter(c => !c.is_deleted)
    .sort((a, b) => String(b.created_at).localeCompare(String(a.created_at)))
    .map(c => ({
      at: c.created_at, who: c.user?.handle ?? null, msg: c.message,
      reply_to: c.parent_id ? (byId[c.parent_id]?.message ?? null) : null,
      resolved: !!c.resolved_at
    }));
})()
""")

The file key is the middle segment of any share URL: figma.com/design/<FILE_KEY>/<slug>.

Response shape

{ meta: [...] } — one flat array, every comment in the file (threads included). Useful fields per comment:

  • message — plain text; message_meta is the structured form
  • user.handle / user.img_url — author display name
  • parent_id — non-null on replies; join against id to rebuild threads
  • created_at, resolved_at, is_deleted
  • client_meta — the pin: node id + offset when anchored to a node
  • thumbnail_url — pre-rendered crop around the pin (commentx/commenty query params carry the pin position)

Traps

  • figma.com/design/<key> URLs auto-launch the desktop app and leave the tab on an interstitial ("Opened ... in Figma app" / "Open here instead"). No need to click through — the interstitial is on www.figma.com, so the comments XHR works right there without ever loading the heavy editor.
  • /api/comments?file_key=<key> does not exist (404). The file-scoped path above is the real one.
  • The current js() helper awaits returned promises and accepts an async function expression.