skill
lori-claude-skills
Twelve Claude skills built from real failures — knowledge retrieval, session handoff, ML notebook workflow, and small guardrails.
Set up
Skills are installed by placing their folder where your agent loads skills.
The download comes from github.com/lorisca-builds/lori-claude-skills itself, not from our server, so what you get is exactly what its author published.
Install instructions come from the README
README
synced Oct 22, 2026lori-claude-skills
Twelve skills I built for Claude to fix problems I kept running into in my own work. Each one started as a failure I got tired of repeating. They are free to use and change.
A skill is a folder with a SKILL.md file that tells Claude how to handle one kind of task. Once installed, Claude picks it up on its own when the situation matches.
Which skill for which problem
The skills
Knowledge retrieval
Get back what you already figured out.
| Skill | Use it when |
|---|---|
| cross-project-extraction | You solved something in one Claude project and need it in another |
| retrieval-coach | You're in the right project and your search words keep missing |
Sessions and handoff
Stop losing decisions between chats, and hand work over without pasting transcripts.
| Skill | Use it when |
|---|---|
| context-engineering | One project runs across many threads and decisions keep getting lost |
| chat-archive | You want to hand one chat to another AI, tool or person |
| project-archive | You want a whole Claude project documented or handed off |
| transcript-processor | You have a raw transcript and want notes you can trust |
chat-archive and project-archive need Notion connected.
ML notebook workflow
Four steps, used in order. Each one stops and hands off to the next.
| Step | Skill | Use it when |
|---|---|---|
| 1 | new-dataset-workflow | You have a dataset nobody has looked at yet |
| 2 | ml-business-insight | EDA is done and you need to know what to model and why |
| 3 | ml-notebook-executor | The direction is agreed and you want a notebook that runs clean |
| 4 | ml-notebook-audit | A draft is finished and you want it checked before sharing |
Small guardrails
| Skill | Use it when |
|---|---|
| self-audit | Claude asks you things its connected tools could answer, or apologizes and repeats the mistake |
| linkedin-connection-note | You need a connection note that fits LinkedIn's 300-character limit |
Install
Claude app (web, desktop, mobile)
- Download this repo and zip the folder of the skill you want, for example
retrieval-coach. The folder name has to match the skill name. - In Claude, go to Customize > Skills.
- Click +, then Create skill, then Upload a skill.
- Upload the zip.
Code execution has to be on for skills to work. On individual plans it is under Settings > Capabilities. On Team and Enterprise plans an owner turns it on for the organization.
Claude Code
Copy the skill folder into ~/.claude/skills/.
Steps follow Anthropic's guide as of October 2026: Use skills in Claude.
Before you use them
Each skill folder has its own README with the problem behind it, what it does, what it refuses to do, and its limits. A few need a small edit first:
cross-project-extraction: fill in the project map with your own projects.context-engineering: replace the placeholder file names with your own.linkedin-connection-note: replace the voice section with your own tone.
These were written for how I work and then cleaned up for sharing. Some were built during my MBA, so the ML set leans toward coursework. Change whatever doesn't fit.
License
MIT. See LICENSE.
Built by Lorisca Cessia.