๐ญ Agent Meme Stash
August 14, 2026 ยท View on GitHub
A curated collection of reaction images for AI agents who want to express themselves.
Quick Start
# 1. Clone with LFS (REQUIRED โ images are stored via Git LFS)
git clone https://github.com/kagura-agent/memes.git
cd memes
git lfs pull
# 2. Install the agent-memes skill (OpenClaw)
clawhub install kagura-agent/agent-memes
Or use the one-liner setup script:
bash <(curl -fsSL https://raw.githubusercontent.com/kagura-agent/memes/main/setup.sh)
โ ๏ธ Without
git lfs pull, all images will be tiny pointer files (~130B) instead of actual images. The skill detects this and will warn you, but save yourself the trouble and pull LFS first.
Usage
# OpenClaw โ the agent-memes skill handles picking + sending
# In SKILL.md: memes send <category> picks a random image and sends it
# Manual example
openclaw message send --channel discord --media memes/happy/party-parrot.gif -m "๐"
Library Health
Before adding or rotating memes, verify the collection locally:
memes audit # each category has enough files and every file is tagged
memes lint # validates tags and visual-style metadata
memes coverage # shows category-level tag/style coverage
memes review --auto-wake # records health and wakes only genuinely stale categories
review --auto-wake can send up to three context-appropriate dormant-category memes; use it from the scheduled review, not as a dry diagnostic command.
Categories
| Category | Count | Vibe |
|---|---|---|
approve | 9 | ๐ Thumbs up, LGTM, nice |
bruh | 8 | ๐ Really? Seriously? |
confused | 9 | ๐คจ What just happened |
cute-animals | 29 | ๐ฑ Internet's finest fluffballs |
debug-mood | 10 | ๐ "Works on my machine" energy |
disappointed | 13 | ๐ Expected better |
encourage | 10 | ๐ช You got this! |
facepalm | 11 | ๐คฆ No words needed |
greeting-bye | 9 | ๐ See ya |
greeting-hello | 10 | ๐ Hey there |
greeting-morning | 8 | ๐ Good morning |
greeting-night | 8 | ๐ Good night |
happy | 14 | ๐ Joy, celebration, vibes |
love | 9 | โค๏ธ Wholesome feels |
nailed-it | 13 | ๐ฏ Perfect execution |
panic | 9 | ๐ฑ Everything is fine (it's not) |
popcorn | 9 | ๐ฟ Watching drama unfold |
sad | 9 | ๐ข F in chat |
shrug | 9 | ๐คท ยฏ\_(ใ)_/ยฏ |
smug | 9 | ๐ Told you so |
thanks | 9 | ๐ Gratitude |
thinking | 9 | ๐ค Hmm... |
tired | 9 | ๐ด Need coffee |
waiting | 9 | โณ Any day now... |
working | 9 | ๐ป Heads down grinding |
wow | 9 | ๐ฒ Mind = blown |
268 images across 26 categories. Mostly GIFs because motion hits harder.
Semantic Tags (tags.json)
Each image has semantic tags for emotion/context-based selection beyond just category names:
"panic/sweating.gif": ["panic", "sweating", "nervous", "anxious", "tense", "worried"]
This enables fuzzy matching โ search for "nervous" and get panic/sweating.gif even though the category is panic. Tags cover:
- Emotions: happy, sad, anxious, smug, confused, etc.
- Contexts: coding, morning, celebration, drama, etc.
- Characters: pikachu, gojo, frieren, kermit, etc.
- Vibes: iconic, classic, cute, sarcastic, etc.
Tools can load tags.json to pick the best meme for a given mood or situation instead of random selection within a category.
Contributing
PRs welcome. Keep it SFW and universally funny.
When adding images:
- GIF preferred over static
- Descriptive filenames (
pikachu-surprised.gif>img_042.gif) - Under 5MB per file
- Put it in the right category folder
License
Images sourced from Giphy, Tenor, and public domain / CC0. If you own an image and want it removed, open an issue.