Claude-Skills
A set of two skills and three Claude Code mod prompts: self-improving skills and shadcn interface design.
- 63 ★ on the repository
- Last commit
Extended access
Why this level
- The skill proposes edits to other skills' files and the mods from the prompts touch deploys and a production database.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/ai-unleashed-claude-skillsDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The skill proposes edits to other skills' files and the mods from the prompts touch deploys and a production database.
This is third-party code. Review the repository files before installing.
What it does
Claude-Skills bundles two skills and three ready prompts for building Claude Code mods. The `autoskill` skill analyzes a past session, finds corrections and repeated user preferences and proposes targeted reversible edits to the skills that were active. Changes are shown for review and applied only after explicit approval. The `frontend-design-shadcn` skill guides distinctive interface creation without template aesthetics: purpose, tone and differentiation first, then working code. The prompts in `mods/` describe `edit-heatmap` with a live map of edits per file, `launch-control` that watches production after a Vercel deploy and `neon-prod-guard` that holds writes to a production Neon database until double confirmation.
Who it is for. Claude Code users with accumulated skills of their own who want corrections from sessions to flow into the instructions.
Good fit when
- After a session with corrections you need to see which became a general rule.
- A repeated correction about code structure goes into the right skill.
- A landing page needs a strong visual idea instead of a stock template.
- Before deploying you build a mod that watches production and wakes the agent on failure.
- Writes to a production database are held until explicit confirmation.
Not a fit when
- A correction concerns one place and will not become a rule.
- The Claude Code version lacks mod support, which the prompts rely on.
- A ready plugin is needed, while the prompts in `mods/` only describe what to build.
Example request
Run autoskill on this session: find corrections worth recording in skills and show edits with confidence levels.Limitations
This is a small personal collection: the README is empty and there is almost no documentation. The `autoskill` skill applies no edits itself and waits for approval, but it proposes changes to skill files, so review each edit. The prompts in `mods/` make Claude build a mod and are not ready modules, and they rely on the plugin-authoring skill and the mods documentation. The Neon database mod and the production watch mod work with live data and should be tested on a test project.
How to disable. Delete the `autoskill` and `frontend-design-shadcn` skill folders from the skills directory. Mods built from the prompts are removed like any plugin.
FAQ
What does autoskill count as valuable?
Direct corrections, repeated remarks and approvals. It skips general practices like DRY and keeps only what is project-specific.
How does autoskill show edits?
As a summary by HIGH and MEDIUM confidence with a signal quote, old and new text and rationale. Applying happens only after a y, n or selective answer.
What does the edit-heatmap mod do?
It watches Edit, Write, MultiEdit and NotebookEdit calls and shows where the agent thrashes, counting reverts.
How does neon-prod-guard work?
It holds a call that changes the production Neon database, shows the changes, asks for double confirmation, snapshots a branch and treats an unknown target as production.
Is there a license?
`autoskill` and `frontend-design-shadcn` carry a license file next to the skill, and the repository has no overall license.
Discussion
No questions or reviews yet.
Related
SkillsSkills, mods, subagents and hooks for Claude Code: the agent inspects your setup and installs what fits after your approval.
claude-mods adds 112 skills, 3 agents and 3 commands to Claude Code: stacks, agent operation, defense, session continuity.
Five skills by Dan McAteer: Claude Code mods, supervised Codex workflows, task handoff and deeper prompts.
11 skills for Claude Code: SKILL.md authoring, prompt tuning, refactoring, change review and planning in files.
A working folder for Claude Code where the agent asks questions and writes a new skill by Anthropic's guidance, not from memory.
A set of skills, slash commands and agents for Claude Code that helps you build your own skills, agents, hooks, slash commands and prompts
README in short
Five instruction files: two skills and three prompts. The author left no overall description, and the structure is visible in the directories.