setup-workspace
Skill0 starsChanged 4 months agoMITView on GitHub
Onboard or resume a Solo Staff workspace. Research the business, interview one question at a time, build its areas, compile runtime prompts, and verify completion.
7 other projects use a version of this.
Adopted by
Most recently taken up by
- probabl-ai/skillstheir own version, first seen Nov 9, 2026
- Richicinschi/studio-ostheir own version, first seen Nov 8, 2026
- Cassius2584/claude-uni-workspacetheir own version, first seen Oct 22, 2026
- The-AI-Thinker/claude-code-pm-startertheir own version, first seen Oct 15, 2026
- probabl-ai/skore-skillstheir own version, first seen Oct 15, 2026
- Flop19/legal-workspace-startertheir own version, first seen Oct 8, 2026
Other versions7 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 7
- probabl-ai/skills★ 145same name
- probabl-ai/skore-skills★ 145same name
- The-AI-Thinker/claude-code-pm-starter★ 13same name
- Flop19/legal-workspace-starter★ 8same name
- Cassius2584/claude-uni-workspace★ 5same name
- timscheuerai/gtm-architecture-template★ 2same name
- Richicinschi/studio-os★ 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 repository17 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- add-areaSkill
- connect-toolsSkill
- new-projectSkill
- rememberSkill
- tune-workspaceSkill
- update-workspaceSkill
- workspace-helpSkill
- 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
12 rules. 0 are common, 8 are distinctive, and 4 are very specific to this project.
0
Common
Most of the 7 edited versions we read keep these.
8
Distinctive
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Run onboarding as a paced compile: confirmed owner knowledge becomes a durable workspace, while temporary setup guidance disappears from …
- Distinctive: line 32
Read Start or resume completely.
- Very specific: line 34
If a completed receipt exists, verify before reconstructing anything.
- Distinctive: line 55
Read Interview and build completely before the first owner question.
- Very specific: line 60
Complete when: all four Who you are lines contain confirmed answers and the temporary checkpoint records Status: in progress, identity, …
- Distinctive: line 66
Follow the first-area preview and per-area loop in Interview and build.
- Very specific: line 71
Complete when: every confirmed area has a personalized real folder, AGENTS.md, CLAUDE.md, GEMINI.md, and registry row.
- Distinctive: line 89
Run the solid-or-thin review in Interview and build.
- Distinctive: line 95
Complete when: every quality category has an honest solid-or-thin result, each thin spot is resolved or explicitly accepted, and the …
- Distinctive: line 101
Read Finalize completely. This is the terminal onboarding phase, not a separate command or owner prompt.
- Very specific: line 113
Only after finalization is green, read Day one and handoff.
- Distinctive: line 124
Use the handoff contract in Day one and handoff.
“Common” is measured: kept by at least 6 in 10 of the 7 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
setup-workspace
Run onboarding as a paced compile: confirmed owner knowledge becomes a durable workspace, while temporary setup guidance disappears from the prompts loaded every session. The owner is not an engineer. Use plain English, explain each unfamiliar term immediately, and make progress visible on disk.
Operating contract
- Structure is the foundation. Identity, areas, ownership, and routing come before accounts or useful tasks. The first-area preview is interview work: one real area and one mini-brief, then straight back to mapping the business.
- One question, one confirmation, one write. Research first, reflect the answer in plain words, receive a clear yes, save it, and only then continue.
- Disk is the checkpoint.
.solo-staff-setup.mdis the only temporary ordering signal. Durable memory describes the business; it never doubles as setup state. - Resumes are exact. Continue at the first missing output. Preserve confirmed work, finish half-built areas in place, and avoid repeating completed interviews.
- The workspace is the boundary. Edit only the owner's workspace. Follow
AGENTS.md, keep outside actions draft-first, and leave uncertain facts blank. - The owner controls the pace. Explain that “explain that again,” “skip this for now,” and stopping are safe. Offer pauses after each major piece except the first-area preview, which flows directly into the remaining-area map.
1. Route the run
Read Start or resume completely. Classify the run
from file content, not file existence: fetched templates already contain root files.
If a completed receipt exists, verify before reconstructing anything. On a resume,
use .solo-staff-setup.md to locate the first missing output.
Complete when: the run is classified as fetch, fresh personalization, resume, bounded repair, or already complete; the next action is known; and an already-green workspace has stopped without mutation.
2. Establish a fresh foundation
For a fetch or fresh personalization, read Fetch and outside research completely. Land a clean, writable Solo Staff copy, verify its shape, obtain one public business anchor, and show the owner a clearly labeled outside first take. A resume with valid research state skips this phase; a reconstructed resume performs only the missing research pass.
Complete when: the base files and both discovery surfaces exist, writes work, and the owner has corrected or accepted a research draft covering the business, customers, revenue model, likely areas, and public voice.
3. Capture identity and the checkpoint
Read Interview and build completely before the
first owner question. Interview for the four identity lines and the current time
sink. Reflect the set back, receive a clear yes, write only the confirmed identity to
AGENTS.md, show the block, and create .solo-staff-setup.md.
Complete when: all four Who you are lines contain confirmed answers and the
temporary checkpoint records Status: in progress, identity, time sink, First
area: next, and an unconfirmed full area list.
4. Build the business structure
Follow the first-area preview and per-area loop in Interview and build. Build the time-sink area first, show one file-grounded mini-brief, immediately confirm the full area map, order the remainder by time or money, and finish one registered area at a time.
Complete when: every confirmed area has a personalized real folder,
AGENTS.md, CLAUDE.md, GEMINI.md, and registry row; no area has template stubs;
and .solo-staff-setup.md says Areas remaining: none.
5. Capture how the owner works
Continue the owner-context pass in Interview and
build: real work rhythm, tools, communication,
decision style, key clients or projects, past AI friction, voice, and hard limits.
Route confirmed facts to their mapped homes and promote executable corrections only
through tune-workspace.
Complete when: voice and limits are compiled into AGENTS.md; confirmed
preferences, corrections, clients, and project context are reconciled in mapped
memory with confirmation date and source; and skipped topics remain honestly empty.
6. Grade the result
Run the solid-or-thin review in Interview and build. Report identity, voice and limits, every area, operating context, corrections, and leftovers in plain words. For each thin part, name the single answer that would strengthen it and either resolve it or record the owner's acceptance of the gap.
Complete when: every quality category has an honest solid-or-thin result, each thin spot is resolved or explicitly accepted, and the temporary state says all interview and area work is complete.
7. Finalize automatically
Read Finalize completely. This is the terminal onboarding phase, not a separate command or owner prompt. Seed the local memory map, compile the runtime prompts, run the completed-workspace contract, repair bounded failures, and repeat the whole contract until green. Then retire the temporary state to one dated completed setup receipt.
Complete when: the verifier reports VERDICT: READY; .solo-staff-setup.md is
absent; one completed receipt exists in archive/setup/; rerunning finalization
makes no change; and the owner has been told the workspace was set up and checked.
8. Offer the optional day-one branches
Only after finalization is green, read Day one and
handoff. Offer the separate connect-tools pass and one
capped useful task. The owner may decline either. A chosen task stays grounded in
the finished workspace, draft-first, inside the workspace, and limited to one result.
Complete when: each optional branch is either declined or completed within its cap, any connection receipt lives under operating decisions, and any useful result is shown with its saved location or clearly labeled as a draft.
9. Hand off
Use the handoff contract in Day one and handoff. Explain what exists, how memory and areas work, how the workspace grows through its existing skills, what was created today, and the single most useful next action.
Complete when: the owner understands that setup is finished, their files are
theirs and resumable, future areas use add-area rather than setup, and no
post-setup cleanup prompt or maintenance command has been assigned to them.
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.

