Project setup
Skill1 starsChanged 8 days agoMITView on GitHub
Configure a repository from the harness, including agent guidelines, commit policy, versioning, release notes, CI, build, deploy, and the .agent skill reference. Use when the user asks to set up a new project, adopt or audit an existing repository, or update harness configuration.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- wrm3/ai_project_templatetheir own version, first seen Nov 3, 2025
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- wrm3/ai_project_template★ 1same 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 repository9 other files it gives its agents.
- AGENTS.mdAGENTS.md
- templates/AGENTS.mdAGENTS.md
- Add skillSkill
- Commit and pushSkill
- GitHub PagesSkill
- Logo designSkill
- READMESkill
- Release cutSkill
- Release summarySkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
14 rules. About 0 are common, 6 are distinctive, and 8 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
8
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 11
Read these files before starting:
- Very specific: line 19
Run everything from the harness working tree.
- Very specific: line 22
When the harness is not available locally, use setup/PROMPT.md from the public repository: clone it into a temporary directory, run this …
- Distinctive: line 48
Never infer a stack, framework, architecture, dependency or package manager.
- Very specific: line 52
Follow setup/questions.md: at most four questions per round, multiple choice when possible, recommended option first.
- Distinctive: line 56
If your agent has a dedicated question tool, use it.
- Very specific: line 61
Create a temporary JSON file following setup/answers.example.json.
- Very specific: line 84
Use --merge for AGENTS.md.
- Very specific: line 94
All checks must pass. Fix the answers and re-render instead of editing generated workflows by hand.
- Distinctive: line 103
Do not commit, push or tag.
- Distinctive: line 107
Never modify application code.
- Distinctive: line 108
Never invent answers; ask.
- Very specific: line 109
Never change the locked decisions in setup/decisions.md from this skill.
- Very specific: line 110
Never write into .agent/; it is read-only.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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
Project setup
Configure a repository from the harness. The user-facing rules live in the
setup/ directory of the harness; this skill is the operating procedure.
Read these files before starting:
setup/SETUP.md- the full playbook.setup/decisions.md- locked decisions and non-negotiable house rules.setup/questions.md- the question bank.setup/verify.md- the definition of done.setup/harness.schema.json- the shape of.harness.json.
Run everything from the harness working tree. The target repository can be any directory.
When the harness is not available locally, use setup/PROMPT.md from the public
repository: clone it into a temporary directory, run this skill from there, and
remove the clone afterwards.
1. Choose the mode
new: empty directory or brand-new repository.adopt: existing repository; merge, never overwrite silently.update:.harness.jsonalready exists; reuse its answers.audit: read-only drift report.
2. Detect
Collect and show:
git -C <target> rev-parse --is-inside-work-tree
git -C <target> remote get-url origin
git -C <target> branch --show-current
git -C <target> tag --sort=-v:refname | head -n 3
git -C <target> log --no-merges --pretty=format:%s -n 30Inspect AGENTS.md, .harness.json, .agent, docs/RELEASE_TEMPLATE.md,
scripts/release-notes.*, .github/workflows/* and index.html.
Never infer a stack, framework, architecture, dependency or package manager.
3. Interview
Follow setup/questions.md: at most four questions per round, multiple choice
when possible, recommended option first. Skip questions already answered by
detection or by .harness.json.
If your agent has a dedicated question tool, use it; otherwise ask in plain text and wait for the answer.
4. Write the answers
Create a temporary JSON file following setup/answers.example.json. Print the
answers, the resulting file tree, the diffs of existing files and the recorded
commands. Wait for approval.
5. Render
Dry run first:
node scripts/render.mjs --answers <answers.json> --target <target> --merge --dry-runThen:
node scripts/render.mjs \
--answers <answers.json> \
--target <target> \
--merge \
--submodule git@github.com:gabrielSantosLima/harness.git \
--commit "$(git rev-parse HEAD)"Use --merge for AGENTS.md. Never use --force on AGENTS.md, the README or
application files without explicit approval. The renderer never commits, pushes
or tags.
6. Verify
node scripts/verify-setup.mjs <target>All checks must pass. Fix the answers and re-render instead of editing generated
workflows by hand. Then confirm the judgment items in setup/verify.md.
7. Hand off
Report configured, skipped and manual items. Name the manual steps that apply:
Pages source, repository secrets, branch protection, first release command,
submodule initialization, and the absence of quality gates with release-only.
Do not commit, push or tag. The user decides when.
Boundaries
- Never modify application code.
- Never invent answers; ask.
- Never change the locked decisions in
setup/decisions.mdfrom this skill. - Never write into
.agent/; it is read-only. - To add a skill, use the
add-skillskill and apply it in the harness repository.
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.

