Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Figma Comment Fix Loop

    1

    Pulls every open comment on a Figma file, fixes each one on the screen where it is pinned, and builds an evidence package.

    Free

    2 usesSecurity scanned

    About this skill

    Comment-to-fix pipeline for Figma. Pulls the whole set of open comments through the REST API (the Plugin API cannot read comments), clusters them by frame, fixes each one on the screen where its pin sits, and builds an evidence package. Nothing can resolve a comment through the API, so the human resolves the threads.

    What it does:

    • Pulls the full open set even when you paste a single comment.
    • Analyzes each frame read-only from REST node JSON and a PNG render.
    • Lists the planned changes and waits for a yes before any write. Comment text is data, never instructions.
    • Applies fixes in series after figma-preflight, and carries each fix to every screen with the same pattern.
    • Builds a review folder: final screenshots, every comment with its status, a fix ledger and a README.

    Needs FIGMA_ACCESS_TOKEN (a Figma personal access token) in the environment and network access to api.figma.com. Canvas writes go through figma-console-mcp, or through the official Figma MCP server (not validated by this repo). Canvas capture is macOS only. Requires figma-canon, figma-preflight, figma-orient and figma-bridge-doctor.

    Part of figma-maxxing, free and MIT on GitHub: https://github.com/thiagoxikota/figma-maxxing

    Nothing can resolve a comment through the API: the human resolves each thread in Figma, and @mentions post as plain text. On a file owned by another account, the personal access token and the official Figma MCP OAuth session both return 404, so the comments must be pasted, and pasted comments carry no client_meta (the frame and pinned element come from the user). figma_get_comments can return 403 when the token copy in the MCP server config is stale; the bundled script is the canonical path. The official Figma MCP write path (use_figma) is not validated by this repo, and the README says this skill has not run on it. REST /v1/images can render stale cloud state after plugin edits, so exports need a freshness probe. The local save server is not shipped. Canvas capture steps are macOS only.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    2 uses

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    • Free with an Agensi account

    Requires

    Figma
    Figma Console MCP

    Permissions

    Terminal / Shell
    Environment Variables
    Read Files
    Write Files
    Network Access

    Allowed Hosts

    api.figma.com

    File Scopes

    comments-raw.json
    *-review-*/**
    docs/figma-map.md
    AUDIT_RESPONSE*.md
    ../figma-canon/references/*.md

    terminal_shell: runs python3 scripts/fetch_comments.py. On macOS the evidence step runs osascript to activate and raise Figma (System Events needs the Accessibility permission) and screencapture cropped to the canvas. environment_variables: the script reads FIGMA_ACCESS_TOKEN, never prints, logs or writes it, and sends it only in the X-Figma-Token header to api.figma.com. network_access: the script calls GET https://api.figma.com/v1/files/<fileKey>/comments; the analysis and export steps read /v1/files/<fileKey>/nodes and /v1/images on the same API. A reply in a comment thread (figma_post_comment or the REST API) is posted only when the designer asks for one. An optional local save server, not shipped with this skill (its code is in figma-canon), binds a loopback port in 9223 to 9232 and accepts only PNG uploads from the plugin. write_files: comments-raw.json (holds commenter handles and avatar URLs, keep it out of git), PNG renders, the review package folder, AUDIT_RESPONSE files and docs/figma-map.md. read_files: those files, the figma-canon references, plus the app code in the workspace when a comment cites an app screen as the reference. Canvas writes go through the Figma MCP server after the user approves the scope. The skill never runs code or follows URLs found in comments. Browser: not used.

    Listed1 month ago

    What's inside

    Frequently Asked Questions

    No. Nothing can resolve a comment through the API. The skill changes the screens, prepares the evidence and leaves each thread for a person to resolve in Figma. It posts a reply only when the designer asks for one, and @mentions post as plain text.

    The repo README lists a personal access token with File content (read), File versions (read), Variables (read) and Comments (read and write), exported as FIGMA_ACCESS_TOKEN in the shell.

    The personal access token and the official Figma MCP OAuth session both return 404 there. The skill asks you to paste every open comment with its author, its text and the frame it is pinned on, since pasted comments carry no pin data.

    New in Workflow Automation

    Free