🪴 Brain

Slack GIF Creator

3 min read

Slack GIF Creator

🗒️ slack-gif-creator is an official skill in anthropics/skills (Apache-2.0) that builds animated GIFs for Slack. It is built on Pillow, imageio and numpy and ships a GIFBuilder, validators (is_slack_ready), easing functions and a frame composer. It targets 480x480 message GIFs and 128x128 emoji at 10-30 fps. In the Meme Generation for Coding Agents research it was the best first-party starting point for animated reaction memes, not for captioned templates.

Description

  • Tested by the research agent: a 480x480, 20-frame GIF built in 0.66 s, was 372 KB and passed is_slack_ready.
  • Polish needs a font. The bundled draw_text uses Pillow’s default 10 px font, which draws every Polish diacritic (including ó) as an empty box. With a TTF (Impact, Arial or Anton) Polish renders correctly.
  • Windows crash risk. GIFBuilder.save() prints a check mark after writing the file, so with cp1250 stdout it should raise UnicodeEncodeError unless PYTHONUTF8=1 is set. Inferred from a test of the print, not from running save() (imageio was not installed).
  • Known bugs, none fixed upstream: validators check dimensions only, not the 128 KB / 50-frame limits Slack requires (#116, open since 2025-11-19); delays are truncated to whole centiseconds so GIFs play faster than requested (#1870); validate_gif computes duration from the last frame’s delay (#1811); about a dozen reports and fix PRs are open, none merged as of 2026-10-15.
  • Adoption (registry data): skills.sh shows 73K installs and three passing audits, which I treat as a signal, not proof.

Download or use

/plugin marketplace add anthropics/skills then /plugin install example-skills@anthropic-agent-skills (installs 12 example skills; issues #189 and #919 report duplicate content with document-skills), or npx skills add https://github.com/anthropics/skills --skill slack-gif-creator, or copy the folder into ~/.claude/skills/. Python 3.10 or newer and pip install pillow imageio numpy; avoid pip install -r requirements.txt, which also pulls imageio-ffmpeg and its bundled ffmpeg .exe, which GIF output does not need.

Reasoning for

Animation is the one place where a ready first-party skill beats my own code. I would use it as the GIF layer behind the caption renderer, always with a bundled TTF and PYTHONUTF8=1, and add my own size and frame-count check on top of its validators. For plain captioned frames, ffmpeg with palettegen and an .ass subtitle file was faster to wire (a captioned GIF to GIF conversion took 0.8 s). Smart App Control risk is LOW on the Pillow route.

Alternatives considered

  • ffmpeg with fontfile= or ASS subtitles: Polish-correct, fast enough for a hook, but drawtext does not wrap text.
  • HyperFrames (installed): real motion design; rendered a 480x480 Polish GIF in 5.3 s, but a root-element background turns transparent in GIF mode and telemetry is on by default.
  • Remotion: needs a company licence at four or more employees.
  • Satori frames + ffmpeg: the route in my own design: 12 frames at 480x480 came to 175 KB with Polish intact.

Resources

Template: tool

Graph View

Backlinks