From Latin elucidare — to bring into clear light
Elucid makes Claude honest, grounded, and educational by default — for everyone, on every topic.
When you use Claude normally, it has some annoying habits:
- Says "Great question!" before every answer (fake)
- Answers confidently even when it's guessing or using old information
- Uses technical words without explaining them
- Dumps a wall of text instead of teaching step by step
- Adds unnecessary warnings to everything
Elucid fixes all of this — permanently, for any topic — without you having to ask every single conversation.
Before Elucid:
"Great question! I'd be happy to help. The reason your phone is crashing is likely due to insufficient RAM. Random Access Memory, commonly referred to as RAM, is a type of volatile memory..."
After Elucid:
"Your phone is crashing because it's running out of RAM — the temporary memory apps use while running. Here's what's happening and how to fix it..."
Same information. No fake warmth. Jargon explained naturally. Gets to the point.
1. Teaches while answering Before every step or command, it explains what it is and why you're doing it. After every result, it explains what it means. Works for any topic — not just coding.
2. Checks facts live For anything that might be outdated (news, new products, current events, AI models), it searches live instead of guessing from old data. It clearly says "I think" vs "I know."
3. Removes the fluff No fake enthusiasm. No repeating your question back to you. No unnecessary warnings. Just clear, direct explanations.
npx skills add ypatole035-ai/ElucidStep 1 — Create the skills folder on your device:
mkdir -p ~/.claude/skills/elucid/referencesWhat this does: creates a folder where Claude looks for skills
Step 2 — Download the skill files:
curl -o ~/.claude/skills/elucid/SKILL.md \
https://raw.githubusercontent.com/ypatole035-ai/Elucid/main/SKILL.md
curl -o ~/.claude/skills/elucid/references/examples.md \
https://raw.githubusercontent.com/ypatole035-ai/Elucid/main/references/examples.mdWhat this does: downloads the instruction files that teach Claude how to behave
Step 3 — Restart Claude Code if it's running.
That's it. Elucid is now active.
Copy the contents of SKILL.md and paste it as a custom system prompt in Claude.ai settings. Elucid will be active for all your conversations.
Start a new conversation with Claude and ask any question. Notice:
- No "Great question!" at the start
- Technical terms get explained naturally
- Answers are direct and educational
Or test directly:
"What is RAM?"
Without Elucid, Claude starts with pleasantries and dumps a definition. With Elucid, Claude explains it like a teacher using a real analogy.
If you're using Claude Code inside Termux on Android:
# Create the skills folder
mkdir -p ~/.claude/skills/elucid/references
# Download the files
curl -o ~/.claude/skills/elucid/SKILL.md \
https://raw.githubusercontent.com/ypatole035-ai/Elucid/main/SKILL.md
curl -o ~/.claude/skills/elucid/references/examples.md \
https://raw.githubusercontent.com/ypatole035-ai/Elucid/main/references/examples.mdThen restart Claude Code. Done.
Elucid doesn't seem to be working Make sure the files are in the right location:
ls ~/.claude/skills/elucid/You should see SKILL.md listed. If not, repeat Step 2 of installation.
Claude still says "Great question!" Try starting a fresh conversation. Skills activate at the start of sessions.
I want to turn it off for one conversation Just say "ignore Elucid for this conversation" at the start. Claude will respond normally.
Does this work for all topics? Yes. Coding, news, history, phones, science, AI — anything.
Does this make Claude slower? No. It changes how Claude writes responses, not how fast it thinks.
Does this work on my phone? Yes, including via Termux on Android.
Is this free? Yes. MIT license. Free forever.
Can I contribute? Yes! See CONTRIBUTING.md. You don't need coding skills — you can contribute before/after examples, test the skill on your device, or translate the README.
Elucid was built because most AI skills are made for developers. This one is made for everyone else — the self-learner, the curious person, the student who wants to understand what's happening, not just follow instructions.
The name comes from the Latin elucidare — to bring something out of darkness into clear light. That's the goal.
Companion project: llamdrop — run local AI models on any Android phone, no PC required.
License: MIT — free to use, modify, share Maintainer: DeVen + community Version: 0.1 README.md