tour
Skill0 starsChanged 20 days agoMITView on GitHub
Walk the founder through how the kit works, one piece at a time - the loop, the two canvases, a test card, a running test's evidence log, a closed test's learning card, the board and the dashboard. Uses the Bean Box example until the kit is set up, then the founder's own files. Use when the founder types /tour or says "show me how this works", "walk me through it", "what is all this".
10 other projects use a version of this.
Adopted by
Most recently taken up by
- ngscheurich/skillstheir own version, first seen Oct 20, 2026
- parcadei/Continuous-Claudetheir own version, first seen Oct 17, 2026
- ClarvAI/job-search-hqtheir own version, first seen Oct 15, 2026
- parcadei/Continuous-Claude-v2their own version, first seen Oct 8, 2026
- zvadaadam/az-skillstheir own version, first seen Sep 28, 2026
- drafthq/drafttheir own version, first seen Jul 30, 2026
Other versions10 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 10
- parcadei/Continuous-Claude★ 3.9ksame name
- parcadei/Continuous-Claude-v2★ 3.9ksame name
- parcadei/Continuous-Claude-v3★ 3.9ksame name
- zvadaadam/az-skills★ 50same name
- drafthq/draft★ 40same name
- allierays/agentic-loop★ 38same name
- YehudaFrankel/clankbrain★ 16same name
- seandavi/scriptorium★ 5same name
- ngscheurich/skills★ 1same name
- ClarvAI/job-search-hq★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository6 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- get-startedSkill
- log-experimentSkill
- new-experimentSkill
- pull-evidenceSkill
- sync-dashboardSkill
The file
We split this skill into the rules it gives an agent
2 rules. 0 are common, 0 are distinctive, and 2 are very specific to this project.
0
Common
Most of the 10 edited versions we read keep these.
0
Distinctive
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 12
Which files: If kit.config.json's business ends with (example), tour the Bean Box example and say once, at the start, that it is a made-up …
- Very specific: line 48
If the kit still holds the example, close with one line: "/get-started sets up your own business: about fifteen minutes of questions, and …
“Common” is measured: kept by at least 6 in 10 of the 10 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
A guided look at a working loop, about five minutes. Six stops. At each stop: read the file, show a few real lines from it (a header excerpt or three table rows, never the whole file), say in two or three plain sentences what it is for, and name the phrase the founder says to Claude to make or change it. Then ask "next?" and wait: they may have a question first. Keep each stop under fifteen lines in chat.
Which files. If kit.config.json's business ends with (example), tour the Bean Box example
and say once, at the start, that it is a made-up coffee subscription. Otherwise tour the founder's
own files, and skip a stop with nothing to show yet (say so in one line: "no test has closed yet, so
there is no learning card to show").
The stops
- The loop, and who does what. Show the loop (hypothesize → experiment → learn → decide → one next hypothesis). The founder decides when a test goes live, whether a bar is right, and every persevere, pivot or kill. Claude writes the cards, reads the numbers and keeps the board and the dashboard current.
- The two canvases. From
experiments/canvas/VP-A.md, show theJ1,P1andG1rows with their Tag, Matters and Evidence. Matters is how much the business depends on the box; Evidence is how much is known. The riskiest box (riskiestin the header) is high Matters, low Evidence: that is what to test next. Then one line onLC-A.md: the whole business model in nine blocks, and a block is only "covered" once a row reaches Evidence 2. - A test card. From the running experiment's
test-card.md(in the example,0002), show the header linestitle,method,boxes,barandstatus, then the "How we measure" table's first row. One small claim, the cheapest believable method, a bar set before it starts, and the measurement planned before anything goes live. Phrase: "test the idea that …". - While it runs. From its
daily-evidence-log.md, show the latest entry's first lines. A read is dated and compared with the bar, and an interim read never decides anything. Phrase: "pull the numbers for 0002". - Closing a test. From a closed experiment's
learning-card.md(in the example,0001), show the observation, the decision and the canvas change. The result is read against the bar, the founder makes the call, the canvas box moves (in the example,J1from belief toseen, Evidence 2), and Claude proposes exactly one next hypothesis. Phrase: "log 0001". - The board and the dashboard. Show the first two rows of
experiments/PROGRESS.md: it is built from the card headers and never edited by hand. The dashboard is a private page on claude.ai with four tabs: Today (what is running and what waits on the founder), Board, Model (the riskiest boxes and the assumptions map) and Log. Ifkit.config.jsonhas adashboard_url, give the link. If it is empty, offer to create it now with/sync-dashboardso they can see it on their phone;/get-startedreuses the same dashboard later.
The end
If the kit still holds the example, close with one line: "/get-started sets up your own business:
about fifteen minutes of questions, and Claude writes the files." If /get-started sent the founder here,
go back to its step 2.
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.

