Building MyLittleJam: practical AI workflows, skills & tools. Sharing real experiments & open-source releases. with claude code mostly #AIWorkflows #AgentSkills
ClawVerse.
Joined August 2022
- Tweets3.1K
- Following1.7K
- Followers477
- Likes13.5K
I ran a free AI SEO check on 13 famous websites.
The SEO tool companies failed their own game:
Ahrefs 4/8
Moz 4/8
Semrush 5/8
None of the three serves markdown to AI agents. Ahrefs and Moz fail the llms.txt check.
The 8-point test, and how to pass it 🧵
🤖 Made with AI
Free and open source (MIT). Works as a Claude Code / Codex agent skill:
github.com/51aitech/llms-txt…
Clone, run it on your site, fix, run again. Done = every row PASS on the live site, not in your repo.
#AISEO #llmstxt
🤖 Made with AI
3 new skills nobody is talking about yet:
1. oil-ui (2.6k): explores 3 UI directions side by side before coding
2. tstack by @theo (2.6k): "full send" takes a change to a merged PR
3. bama by @daveebbelaar (1.4k): Build Any Mac App
Links in reply. #ClaudeSkills
Follow @51AITech
🤖 Made with AI
You built it with AI. What happens at checkout?
floorplan by @haydenschmitty (2.3k stars): a plain-English map of your repo plus a health check for keys in code and endpoints with no sign-in.
github.com/haydenschmitty/fl…
#ClaudeCode #AgentSkills #ClaudeSkills
Follow @51AITech
🤖 Made with AI
Boxes and arrows explain nothing. This does.
anatomy by @wheresryan22 (24k stars): Claude builds your idea as an isometric machine you can take apart. Now with exploded views. No API keys.
#ClaudeCode #AgentSkills #ClaudeSkills
Follow @51AITech
teapawt.org/wheresryan22/status/21…
Your AI video looks like every other AI video.
lemo-opuscar (11k stars): 66 film styles as code for Claude Code, chalkboard, transit map, diorama, or your own from one reference clip.
github.com/lemomo-ai/lemo-op…
#ClaudeCode #AgentSkills #ClaudeSkills
Follow @51AITech
🤖 Made with AI
My AI agent wrote more disclaimers than answers.
Codex on max reasoning kept ending replies with "not verified", "no claim is made", "pending confirmation".
3 lines fixed it. Copy them below 🧵👇
🔖 Bookmark it. ➕ Follow @51AITech for agent fixes that work.
#Codex #AIAgents
ALT Illustrated cover: a woman in a blazer stares at her laptop, jaw dropped, spilling coffee. She asked her AI "Did you fix it?" and it replied only with disclaimers: Not verified. No claim is made. Not tested. Pending confirmation.
The prompt. Paste into your AGENTS.md:
"Report in max 3 lines: what you did + the evidence. Don't list what you didn't do or didn't claim. No disclaimers. If something wasn't done, say it once, plainly."
And in config.toml:
model_reasoning_effort = "medium"
After the change, I asked it a real question. One clean answer, zero hedging. Small test, not a benchmark.
🔖 Bookmark it. 🔁 Send it to the friend whose agent hedges more than it ships.
➕ Follow for the agent rules that actually work.
#Codex #AIAgents
ALT Terminal screenshot of a Codex run on medium reasoning after the change. Asked a real question, it returns one direct numbered answer with no disclaimer lines.
One person is now a 7-person team.
Claude Code can design, build pages, make videos, map architecture, research, cut promos and run your Instagram. 7 free open-source skills, 540k+ stars 🧵👇
✦ Follow @51AITech for more free AI tools
#ClaudeCode #AgentSkills #ClaudeSkills
🤖 Made with AI
7/ Instagram → instagram-agent-skill (4.4k ⭐)
13 skills that run an IG account: Reels hooks, viral research, caption linter.
github.com/Jakeschincariol/i…
🤖 Made with AI
Bookmark this so you have it when you need it.
We test a new AI tool every week → follow @51AITech
🤖 Made with AI