agentleFS
dtiger1889-opsdtiger1889-ops/claude-harness-toolbox/skills/new-project/SKILL.md

new-project

Skill31 starsChanged 32 days agoMITView on GitHub

Scaffolds a new project folder from the harness templates and -- the load-bearing half -- registers it everywhere it must appear (the project ledger, any container index, its backup lane). Use when the user says "/new-project", "new project", "set up a project for X", or promotes an outgrown scratch folder to a project of its own. Do NOT use for adding a subfolder inside an existing project, or for work in a project that already has its CLAUDE.md and CHECKPOINT.md.

51 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use prd, explain and 8 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.
  • GitHubSkill

    prd

    Generate high-quality Product Requirements Documents (PRDs) for software systems and AI-powered features. Includes executive summaries, user stories, technical specifications, and risk analysis.

    Also in 3 projects that use this

  • sunnyclinches44Skill

    explain

    Explains code using real-world analogies and ASCII diagrams

    Also in 3 projects that use this

  • GoogleSkill

    domain-modeling

    Build and sharpen a project's domain model. Use when the user wants to pin down domain terminology or a ubiquitous language, record an architectural decision, or when another skill needs to maintain the domain model.

    Also in 4 projects that use this

  • SentrySkill

    review

    Fixture review skill.

    Also in 5 projects that use this

  • NVIDIASkill

    research

    Before designing a feature, research how comparable libraries solve the problem, check best practices, known pitfalls, and upstream changes. Use when starting a new feature, investigating an unfamiliar area, or the user asks to research something.

    Also in 5 projects that use this

  • Microsoft AzureSkill

    design

    Design rules and conventions for the API Center Portal. Use when editing detail pages, card components, badges, layout, navigation, sidebars, or headers. Covers badge styling, metadata placement, page navigation patterns, and layout alignment.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 48 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Stop if a repo already exists there and diagnose it instead of initializing over it.

    line 56
  2. Distinctive:

    Check every command's result.

    line 62

“Common” is measured: kept by at least 6 in 10 of the 48 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

/new-project -- scaffold + register a project

Scaffolding tends to get done; registration tends to get skipped, and a project no index names is invisible to every later session. Never end this skill with the folder scaffolded but unlisted. Conventions live in the workspace-level CLAUDE.md; the skeletons are this repo's templates/, copied wherever you keep them.

Step 0 -- the three questions (ask only what the conversation hasn't answered)

  1. Name -- short kebab/snake slug; folder name = project name. Name it for what it DOES, not for the model that built it, so the name survives a change of agent.
  2. Location -- top-level project, or an occupant of a container folder (one for experiments, one for small apps)? Default: experiments and one-offs start in the experiments container and get promoted later; anything with a standing goal goes top-level.
  3. Backup lane (required -- every folder needs one):
    • Synced folder -- if it lands under a folder a sync service covers, say so. A two-way sync copies deletions too, so it is coverage, not a backup.
    • Private git repo -- only when the user asks for git; see Step 1b. A public repo needs the user's explicit per-repo decision and their publish review, never a --public flag by default. Inside an existing repo, the parent repo IS the lane: verify the new folder is not gitignored.

Step 1 -- scaffold from the templates (never freehand)

templates/CLAUDE.md.template      -> <proj>/CLAUDE.md
templates/CHECKPOINT.md.template  -> <proj>/CHECKPOINT.md
templates/gitignore.template      -> <proj>/.gitignore   (new repo only)

Fill every [bracketed placeholder]. CLAUDE.md stays <= 30 lines with a line-2 version stamp (<!-- v1 -- YYYY-MM-DD -->). CHECKPOINT.md gets Last updated: from the real clock in UTC (run a command, never guess) and the sections Status / Goal / Key decisions / Open threads / Files that matter / Next step / Harness changelog. A private repo's .gitignore covers only secrets, personal data and machine junk, never a working file (CHECKPOINT, CLAUDE.md, specs, docs), because a resumed session needs them.

Step 1b -- create a private repo only after the project files are complete

  • Inspect the folder first for nested .git directories, secrets or personal data, generated output, databases, archives and large files; write .gitignore first. Stop if a repo already exists there and diagnose it instead of initializing over it.
  • git init -b main. If the folder sits inside a sync service, keep the git metadata outside it (--separate-git-dir <path outside the synced tree>), because a sync client rewriting a live .git folder corrupts it.
  • Stage only after every placeholder is filled; inspect the staged tree and its largest files before committing. Create the remote private, add origin, push main.
  • Check every command's result; never suppress a failure and print success after it.

Step 2 -- register (do ALL that apply)

  • Top-level project -> a row in the project ledger (the one file listing every project, with a one-line purpose) and a note in the workspace CHECKPOINT changelog.
  • Container occupant -> the container's CLAUDE.md contents list AND its CHECKPOINT.md Status line, same turn.
  • Either way -> name the backup lane in the project's CLAUDE.md or the container's conventions, so a later audit doesn't have to re-derive it.
  • New private repo -> every list that tracks repos (backup script, repo monitor, public/private registry), same turn.

Step 3 -- research-first opening

The project's first working turn writes a ## Research block in its CHECKPOINT (prior art, constraints, open questions, sources) before any implementation. Shipped solutions first: GitHub, published skills, community writeups; default to adopt/adapt with a citation. Building from scratch needs a stated reason.

Step 4 -- verify before closing

Copy this checklist into the reply and tick each item:

- [ ] CLAUDE.md <= 30 lines, CHECKPOINT.md <= 120 lines (counted, not estimated)
- [ ] every index from Step 2 re-read and shows the new name
- [ ] backup lane real: inside a synced tree, repo pushed, or parent repo covers it un-ignored
- [ ] new private repo: branch main, remote private and pushed, on every repo list
- [ ] "registered in: ..." reported to the user

Promotion variant (container -> top-level)

Same skill, reversed entry: move the folder up (a real move, never copy + delete), add the ledger row, remove it from the container's contents list and Status, note the move in both changelogs, and pick a new backup lane the same turn if leaving a git repo's tree took the old one away.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.