setup
Skill6 starsChanged 3 days agoMITView on GitHub
Guided workspace onboarding — interview the founder, fill in data/product.md and the companion trackers (strategy, brand, offer, metrics), prune the seeded venue list to the ICP and re-read the kept venues' posting rules on the web, activate the matching stage plan, delete the example documents, offer git init once, and export the .agents/product-marketing.md bridge file that marketing skill packs read. Use when the user says "run the setup" or "set up the workspace", or when What is it, Ideal customer profile, Positioning or Messaging pillars in data/product.md still shows a bare ✏️ (one marked "✏️ skipped" does not count).
192 other projects use a version of this.
Adopted by
Most recently taken up by
- openstory-so/openstorytheir own version, first seen Nov 16, 2026
- furkandvrc/conductortheir own version, first seen Nov 16, 2026
- futorio/red-greentheir own version, first seen Nov 16, 2026
- JCodesMore/jobs-for-ai-agentstheir own version, first seen Nov 16, 2026
- finolabs/finolabs-vibecodetheir own version, first seen Nov 16, 2026
- vincentventalon/claude-code-video-editing-skilltheir own version, first seen Nov 15, 2026
Used with
Projects with it also use add-atomic-chat-tool, add-dashboard and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
MicrosoftSkilladd-atomic-chat-tool
Add Atomic Chat MCP server so the container agent can call local models served by the Atomic Chat desktop app via its OpenAI-compatible API.
Also in 8 projects that use this
MicrosoftSkilladd-dashboard
Add a monitoring dashboard to NanoClaw. Installs @nanoco/nanoclaw-dashboard and a pusher that sends periodic JSON snapshots.
Also in 8 projects that use this
- +10 more from microsoft
Other versions192 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 192
- anthropics/knowledge-work-plugins★ 36ksame name
- garrytan/gbrain★ 33ksame name
- gavrielc/nanoclaw★ 31ksame name
- nanocoai/nanoclaw★ 31ksame name
- qwibitai/nanoclaw★ 31ksame name
- Q00/ouroboros★ 6.4ksame name
- pytorch/executorch★ 5.2ksame name
- aws/agent-toolkit-for-aws★ 3.5ksame name
- agenticnotetaking/arscontexta★ 3.5ksame name
- spotify/portal-ai-plugins★ 2.8ksame name
182 more not listed
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.
The file
We split this skill into the rules it gives an agent
7 rules. 0 are common, 4 are distinctive, and 3 are very specific to this project.
0
Common
Most of the 169 edited versions we read keep these.
4
Distinctive
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 14
Assess: Read data/product.md, data/strategy.md, data/brand.md, data/offer.md, data/metrics.md.
- Distinctive: line 30
Interview: Ask in batches, conversationally (don't interrogate all at once):
- Very specific: line 127
Delete the examples one at a time.
- Distinctive: line 145
If a line is left that is not the founder's own text, open the file and remove the bullet or sentence by hand.
- Very specific: line 157
Never leave a ✏️ block half-filled — ask follow-ups until each section is concrete.
- Distinctive: line 163
Don't cancel venues silently.
- Distinctive: line 164
Never invent numbers, quotes, or competitor facts — "unknown" beats plausible fiction.
“Common” is measured: kept by at least 6 in 10 of the 169 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
Workspace setup
Turn the blank workspace into this founder's campaign memory. Everything downstream (drafts, venue choices, outreach) reads what you write here — optimize for the founder's own language, not marketing-speak.
Steps
- Assess. Read
data/product.md,data/strategy.md,data/brand.md,data/offer.md,data/metrics.md. List which ✏️ blocks are unfilled; a block marked✏️ skipped <date>was left on purpose — ask about it once, and leave it if the answer is still "later". If everything is filled, ask what to revise instead of re-interviewing. Two one-time notes, neither of which you repeat or act on unasked:- Git. If
git rev-parse --is-inside-work-treefails, the folder is not a repository (the iWe app creates a plain folder). Offergit initonce, so the campaign has a history; run it only on a yes. Without git, skip the commit in the last step. - Skill pack. Mention that a marketing skill pack such as
marketingskills can
add copywriting and launch technique, and that the user installs it
themselves from its own instructions. Never run an installer. If technique
skills beyond this workspace's six are already present, record the pack in
a one-line "Toolkit" note at the bottom of
data/strategy.md.
- Git. If
- Interview. Ask in batches, conversationally (don't interrogate all at
once):
- Product: the party-sentence one-liner; who exactly it's for and where they hang out; who it's NOT for (disqualifiers, anti-persona).
- Alternatives: what people use today; the sharpest difference; the 2–4 competitors worth tracking.
- Evidence: proof points; objections heard so far and the honest responses; verbatim phrases customers use for the problem.
- Money: current tiers/pricing, the free-vs-paid boundary, guarantee.
- Voice: how it should sound, words to ban; existing brand assets and where they live.
- Reality: honest stage (first-10 / first-100 / first-1000); weekly time budget; platforms they're comfortable on; baseline numbers they know offhand.
- Write the docs. Fill
data/product.md(all sections, keeping the founder's vivid phrasing; derive 3–5 pillar slugs and confirm them),data/strategy.md(stage, 2–3 channels justified against the ICP, observable goals, cadence),data/brand.md(voice + description variants — draft the 60-char and 150-word variants from the one-liner and confirm),data/offer.md, anddata/metrics.md(what's known; "unknown" is a valid value). Delete the italic instruction lines as sections fill. Add a dated entry to product.md's Changelog. Create a hub per confirmed pillar atdata/pillars/<slug>.md(type: hub,stage: living,created, a one-line statement of the messaging claim) and inclusion-link each fromdata/pillars.md— a link on its own line. Every future publication gets an inclusion link from its pillar hub, which keeps posts reachable in the tree.data/index.mdalready listsdata/pillars.md; it takes link bullets only, never an inclusion link. - Create competitor stubs and segments. For each competitor named in the
interview, create
data/competitors/<slug>.md(stage: active, resource,relation, one-line stance innotes, and the five body sections with ✏️ where unknown) and link it from the competitors hub; deep profiling stays a backlog task. For each audience segment the founder named, createdata/audiences/<slug>.md(stage: activeorplanned) with the five sections, and link it fromdata/audiences.md. Each proof point the founder can show becomes adata/selling-points/<slug>.mdstub with itsevidence. - Prune venues. The seeded list is a menu, not a plan: nothing on it is
worked until the founder picks it.
Deal platforms first. The fourteen
category: dealsvenues sell lifetime access at a steep one-time discount. Ask one question: is a lifetime deal something this product would ever run? Unless the answer is yes, cancel the whole category in one command and do not raise it again:iwe update --filter '{type: community, category: deals}' \ --set stage=cancelled --set status=deprecated --expect 14The venues'
noteskeep their posting rules; record the reason once, as a decision note indata/notes/(kind: decision,origin: human) linked fromdata/notes.md.- Then the rest. Walk
iwe find --filter '{type: community, stage: planned}' -f jsonagainst the ICP. Propose a shortlist (~10 to keep) and the cancellations with one-line reasons. On confirmation, for each cut venue append the reason tonotesand runiwe update -k <venue key> --set stage=cancelled --set status=deprecated— a cancelled venue isstatus: deprecated(the table inSCHEMA.md). Keys aredata/community-<slug>for standalone sites anddata/communities/reddit/<name>for subreddits. - Refresh what stays. The seeds were researched on their
createddate, and some say "rules unverified" innotes. For each venue kept whosestale_afterhas passed or falls within the next month, or whose rules are unverified, open itsresource, correct the posting rules innotesand the body, and setstale_aftersix months out with averifiedstamp. If the page cannot be reached, leave the date and say so innotes. Cancelled venues are not refreshed.
- Activate the plan. In
data/plan.md, put the matching stage plan under## Now(others under Next/Later in stage order) and set its frontmatterstage: in-progress. The two charts on that page count toward a target of 10; forfirst-100orfirst-1000, change the target in the three places it is written:max(10 - datum.paying_users, 0),domain: [0, 10]andy: {datum: 10}. - Export the bridge file. Marketing skill packs (e.g. marketingskills)
read
.agents/product-marketing.mdbefore asking questions. Generate it: start with the line<!-- Generated from data/product.md — edit there and re-run setup to refresh. -->, then the filled content of product.md (all sections including Changelog), the Voice and Descriptions sections from brand.md, and the Current offer section from offer.md. Regenerate this file whenever those sections materially change. Close the loop. Mark the onboarding tasks done (
fill-product-doc,set-strategy,prune-venues):stage: done,completed: <today>, and move their links indata/backlog.mdto the## Donesection. Then remove the example documents — every*-example.mdunderdata/— in this order, from the workspace root:Drop the list bullets that link to an example, while the links are still there to find (the venue files'
## Postsand## Mentions):grep -rlE -- '-example\.md\)' data | grep -v -- '-example\.md$' \ | xargs perl -0pi -e 's/^- (?:(?!\n- |\n\n|\n#).)*?-example\.md\)(?:(?!\n- |\n\n|\n#).)*\n//gms'Delete the examples one at a time;
iwe deleteremoves the hubs' inclusion links to each. A single--filterdelete of all of them fails part-way, so use the loop:find data -name '*-example.md' | sed 's/\.md$//' | sort \ | while read -r key; do iwe delete "$key" --quiet || echo "FAILED $key"; doneTidy the files that changed, then check that nothing is left. The two checks must print nothing:
iwe normalize find data -name '*-example.md' grep -rniE 'plaininvoice|-example' dataIf a line is left that is not the founder's own text, open the file and remove the bullet or sentence by hand.
Then reset
data/metrics-weekly.csvto its header row (week,visits,signups,activated,paying_users,mrr) plus the founder's real weeks, if they have any — its current rows are dummy numbers.- Validate.
iwe schema validatemust pass. In a git repository, commit with a message likesetup: product/strategy/brand/offer filled, venues pruned to <n>, <stage> plan activated.
Rules
- Never leave a ✏️ block half-filled — ask follow-ups until each section is
concrete. When the user says "skip for now", replace the block with
✏️ skipped <today's date>so the gap stays visible. A skipped section does not trigger setup again and does not block drafting; mention it when the work in front of you needs it. - Pillars are slugs (
lowercase-hyphenated), each with a one-line description. - Don't cancel venues silently; the user confirms the pruning list first.
- Never invent numbers, quotes, or competitor facts — "unknown" beats plausible fiction.
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.

