Claude Project Assistant
Teach your AI
how you work.
Pick a skill. Hand it to Claude. Done.
A library of ready-made abilities for AI projects: tidy file systems, trackers, research methods, legal tools and more. No coding. Every one asks first and writes only what you approve.
An illustration, not a recording.
§ 01
A shelf of abilities for your AI.
Each feature is a short, tested set of instructions for one job. Add the ones you want and your AI does that job your way, every time.
Skill
A skill
A folder of written instructions that teaches Claude one job, like keeping files tidy or checking legal citations.
Claude reads it only when the job comes up, so it never gets in the way.
Connector · MCP
A connector
A plug that lets Claude reach things outside the chat: your files, your email, a spreadsheet.
Connectors give Claude hands. Skills tell it what to do with them.
Project
A project
A workspace where Claude keeps your instructions and files together, so you stop repeating yourself.
This library is built to be added to a project.
Why Claude? These abilities need an AI that can load skills, plug into connectors and work on your files. Claude does all three, which makes it an easy place to try them and see what is possible. Many features also read fine as plain written instructions in any AI that can open a link.
§ 02
Get started in four steps.
You do not need to understand any of the above to begin.
- 01
Pick what you need
Browse the library. If you are new, start with the seven Foundation features. They set up the basics everything else builds on.
- 02
Choose how to add it
Paste a link into your AI chat, or download the skill and add it to a tool that supports skills.
- 03
Say its name
Tell Claude, for example, “Install FTR-15, Project Cache.” It reads the feature and starts.
- 04
Answer, then approve
It looks at what you already have, asks only what it must, and writes nothing until you say yes.
Not sure where to start? Open the Library, tick Add on a few cards, and the page lists them in the right install order with a ready-made message to copy.
§ 03
Three ways in.
Use whichever matches the AI you have.
Read it
Any AI that can open a link.
Paste this into your chat and then name a feature.
Read https://claude.sotekzone.com/ai-catalog.md. When I name a feature by its ID, fetch https://claude.sotekzone.com/id/<ID>.md and walk me through it.Install it as a skill
Claude Code, Linux VMs, apps with a skills upload.
Download a skill folder and place it in .claude/skills/, or upload the zip in your app’s skills settings.
Look before you leap
Not ready to install anything?
Every feature has a plain-words page: what it does, when to use it and what it needs first.
§ 04
What you can do with it.
A few favorites, in plain words. There are 42 in all.
Project Cache
Keeps the files Claude reads at the start of every chat up to date. One staging folder holds the current copies, along with an automatically kept index that tells Claude which file to open and when. To refresh, you drag that one folder into the project.
Folder Structure
A standard folder layout for your project (a drop folder, documents, drafts, research, brand, trackers, archive and more), plus extra folders that fit the type of project you have. It only creates what is missing.
Session Handoff Protocol
When a conversation gets too long, Claude writes a summary document that captures everything important. You paste it into a fresh conversation and carry on where you left off.
Multi-AI Research Synthesis System
A method for research questions that are too important or too tricky to trust to one quick answer. Claude writes a careful research prompt and shows it to you. You run it on several AI services and drop the results in your drop folder, and Claude combines them into one polished master document.
Legal Citation Management System
Tracks every case or statute you plan to cite in one report, with a status on each: still to be verified, verified, flagged or dropped. The aim is to catch AI-invented or misdescribed citations before they reach a filing. Nothing goes out below 'verified'.
Project Audit / Health Check
Say 'run an audit' and get a color-coded, one-page health check of your project: what is solid, what needs attention and what is broken, each with a suggested fix.
Browse by category
§ 05
Questions people ask.
Do I need to know how to code?
No. You talk to Claude in plain language. The features do the technical work and explain what they are doing as they go.
Will it change my files without asking?
No. Each feature looks first, asks only what it must, and writes only what you approve. Nothing installs by itself.
What if I only want one or two features?
Take only those. The library tells you when one needs another first, and the picker adds the required ones for you, in order.
Does it only work with Claude?
Claude is the easiest place to start because it supports skills and connectors. Many features also work as plain written instructions in any AI that can open a link.
How do I know the files are genuine?
Every file is listed with its checksum in the integrity manifest, and every version has release notes saying what changed and why.