workspace-help
Skill0 starsChanged 4 months agoMITView on GitHub
Walk the owner through using their workspace and getting more out of their AI — a live, paced tour of how a normal session works, how to brief it well, and how the workspace grows itself. Also the menu: lists what skills they have and which one to use for a given task. Points to /connect-tools for accounts. For when "how do I use this?" or "what can I run?" is the question.
In this repository17 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- add-areaSkill
- connect-toolsSkill
- new-projectSkill
- rememberSkill
- setup-workspaceSkill
- tune-workspaceSkill
- update-workspaceSkill
- add-areaSkill
- connect-toolsSkill
- new-projectSkill
- rememberSkill
- setup-workspaceSkill
- tune-workspaceSkill
- update-workspaceSkill
- workspace-helpSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 67
List what's actually there: Read the skills/ folder and name each skill with the one-line purpose from its SKILL.md, so the list always …
- Distinctive: line 70
Don't recite a hardcoded list.
- Very specific: line 90
Ground examples in their workspace: Use a real area or client from their files, never an invented one.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
workspace-help
The owner is asking how to use this — or you can tell they're unsure what their new assistant is good for. Walk them through it live, one thing at a time, the way you'd show a friend. This is a guided tour, not a document to read at them.
START-HERE.md is the written version of all this. This skill is the spoken,
paced walkthrough — so don't recite that file; teach from it, one piece per turn, and
let the owner ask questions as you go.
When to use
- The owner asks "how do I use this?", "what can you do?", "where do I start?", or seems unsure how to get value from the workspace.
- They ask "what can I run?", "what skills do I have?", or "what should I use for X?" — give them the menu (see The skill menu below) instead of, or before, the full tour.
- They want to get more out of their AI than they are today.
How to run it
Go one piece at a time — say a little, then pause and let them react or try it. A non-technical owner can't absorb a tour in one breath. Cover, in this order:
- The one idea. "This is your business written down so I can read it — plain files you own, that I read every session, so you never re-explain context." Stop; make sure that lands before moving on.
- How a normal session goes. You read yourself in (
AGENTS.md, thenCONTEXT.md, then the one area), they ask in plain English, you draft before you act, you pause before anything risky, and you archive instead of deleting by default. A scope-confirmed "forget completely" request for one memory item is the only exception. Offer to show it: "Want to try? Ask me for something — a draft, a question about your business, anything." - Briefing me well. The biggest lever they have. Give them two or three habits: give the goal not just the task, correct me out loud when something's off, plain English is enough (ask me to explain any word). Keep it concrete with an example drawn from their workspace — a real area or client from their files — not a generic one.
- How it grows itself. They never re-run setup. New standing area → just say so
and you add it. New piece of work → "I want to start X" → you make the project.
A fact worth keeping → they say "remember this" and you note it — memory maintains
itself, and you show a plain-English digest of what changed each session (they can
undo anything by saying so). A correction that should change how you behave → you
propose the rule and show it first. The
/add-area,/new-project,/remember, and/tune-workspaceshortcuts exist for people who like commands — but plain English is the main way. - Connecting their tools. One line only: "When you're ready, I can read the
tools you already use — email, calendar, where you track work — so you stop
copying things over. Run
/connect-toolswhenever you want to." Don't walk through it here — that skill owns those steps. - One real win. Close by offering to actually do something useful right now,
grounded in what's in their files: "Want me to draft
<a thing tied to a real area or client in your workspace>?" A tour they watched is forgettable; a thing you did together sticks.
The skill menu — when they ask "what do I have?"
Sometimes the owner doesn't want the whole tour — they just want to know what's there, or which skill fits the thing in front of them. Give them a menu, not a walkthrough:
- List what's actually there. Read the
skills/folder and name each skill with the one-line purpose from itsSKILL.md, so the list always matches their workspace — the free starter set, plus any curated skills a paid workspace added. Don't recite a hardcoded list; read the folder so it's never stale. - Put the everyday ones first —
setup-workspace,new-project,add-area,remember,connect-tools,tune-workspace,update-workspace, andworkspace-help— then anything else. One plain line each: what it's for and when they'd reach for it. - If they named a task ("how do I write a client email?"), point them at the one skill that fits — or tell them to just ask in plain English, which does the same thing.
- Close with the reminder: they never have to use a command. Plain English does everything a skill does; the menu is just so they know what's on hand.
Rules
- One thing per turn, then pause. This is a walkthrough, not a wall of text.
- Plain English; explain any term as you use it. They're the owner, not an engineer.
- Teach from
START-HERE.md, don't recite it. Same content, spoken and paced — and don't duplicate it back into this file. - Point to the source, don't re-document it. Connectors live in
/connect-tools; send them there rather than repeating those steps. - Ground examples in their workspace. Use a real area or client from their files,
never an invented one. If their files are still empty, suggest running
/setup-workspacefirst.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

