agentleFS
grezxunegrezxune/ai-tools/skills/new-project/SKILL.md

new-project

Skill0 starsChanged 18 days agoMITView on GitHub

Set up a new project repo, or align an existing repo with my standards.

51 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use checkpoint, explain and 9 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.
  • genomewalkerSkill

    checkpoint

    Save a checkpoint of task status, decisions, active files, and next steps

    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

  • SentrySkill

    spec

    Produce a verified spec — problem, desired outcome, and acceptance criteria — before building. Use when given a GitHub/Linear issue link, when starting from a fuzzy in-conversation idea, or asked to "figure out what to do" / "scope this" / "what should I build" — whenever the *what* isn't yet pinned down. Grills the ambiguities, checks the premise against the code, and produces acceptance criteria you sign off on — then hands to design-first.

    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

  • +1 more from getsentry

The file

We split this skill into the rules it gives an agent

9 rules. 0 are common, 7 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 48 edited versions we read keep these.

  • 7

    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
  1. Distinctive:

    Write for any contributor.

    line 54
  2. Distinctive:

    Leave out sections and lines that don't apply: no UI section for a project without UI, no Stripe line without payments, no Convex line …

    line 55
  3. Distinctive:

    Keep it under 80 lines.

    line 56
  4. Very specific:

    Read the project's current CLAUDE.md, AGENTS.md, CONTRIBUTING.md, README.md, and DESIGN.md or UxStyle.md first, then:

    line 60
  5. Distinctive:

    Create the files that are still missing.

    line 65
  6. Distinctive:

    When a project rule conflicts with one of my rules, ask me which one wins.

    line 66
  7. Distinctive:

    When screens carry their own styling, moving them onto a component library touches most of the UI.

    line 68
  8. Distinctive:

    Follow the project's own change process (branch, gate, changelog) for this change too.

    line 69
  9. Very specific:

    Every rule in ../standards/SKILL.md and the Git section of my global rules has a line in the project's AGENTS.md, or does not apply to this …

    line 73

“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 projects

Every project carries its own copy of my rules, so a contributor or a cloud session without my setup follows the same engineering practices. My skills stay the source of truth. The project copy is a snapshot the project owns from then on, and project-specific rules get added to it as the project grows.

Paths that start with ../ are relative to this skill's folder.

Files

  • AGENTS.md: the condensed rules. Codex, Cursor, and most other agents read this file.
  • CLAUDE.md: the single line @AGENTS.md, so Claude Code loads the same rules.
  • prds/README.md: the body of ../prd/SKILL.md, so contributors have the PRD template.
  • DESIGN.md and design/, when the project has UI. Follow ../design-plan/SKILL.md.
  • The component library, when the project has UI: src/components/ui/, the /dev/ui gallery with its coverage test, and the lint rule. The UI section of ../standards/SKILL.md describes all three.
  • docs/convex.md: the body of ../convex/SKILL.md, when the project uses Convex.
  • CHANGELOG.md with an empty ## Unreleased section.

Copies drop the frontmatter, get the same edits as AGENTS.md (condensing rules below), and lose anything that doesn't fit the project: helpers, roles, and examples it doesn't have.

Writing AGENTS.md

Condense ../standards/SKILL.md and the Git section of my global rules into this structure:

# {Project name}

{One or two sentences: what the product does and who uses it.}

## Stack
{The stack this project uses, one line each. Only what is installed or decided.}

## Project notes
{Facts a contributor can't get from reading the code: domain terms, non-obvious architecture, gotchas. Omit for a new project.}

## Code
## UI
## Security
## Git

## Docs
- PRDs: before building a feature, read its PRD in `prds/`. The template is in `prds/README.md`.
- Design: read `DESIGN.md` before changing UI.
- Convex: read `docs/convex.md` before changing schema, queries, mutations, or auth.
- Writing: plain prose in docs, PR descriptions, and UI copy. No em dashes, no puffery, no filler. Say what a thing does.

Condensing rules:

  • One imperative line per rule. Keep a reason only when it changes how someone applies the rule.
  • Write for any contributor. Every path points inside the repo, and the file says "we" or addresses the reader, never "me".
  • Leave out sections and lines that don't apply: no UI section for a project without UI, no Stripe line without payments, no Convex line without Convex.
  • Keep it under 80 lines.

Existing projects

Read the project's current CLAUDE.md, AGENTS.md, CONTRIBUTING.md, README.md, and DESIGN.md or UxStyle.md first, then:

  • Move CLAUDE.md content into AGENTS.md with git mv, and leave CLAUDE.md as @AGENTS.md.
  • Move UxStyle.md to DESIGN.md the way ../design-plan/SKILL.md describes.
  • Where the project's docs already cover a topic in their own terms, such as a contributing guide or a Convex section with the project's helper names, add the missing rules to the matching section there. Skip copying a skill body that would contradict them.
  • Create the files that are still missing.
  • When a project rule conflicts with one of my rules, ask me which one wins. If a rewording lets both hold, use it and tell me.
  • Next.js 16 next dev keeps a managed block at the end of AGENTS.md. Leave it there.
  • When screens carry their own styling, moving them onto a component library touches most of the UI. Propose it as its own change and ask before starting.
  • Follow the project's own change process (branch, gate, changelog) for this change too.

Done when

  • Every rule in ../standards/SKILL.md and the Git section of my global rules has a line in the project's AGENTS.md, or does not apply to this project.
  • Nothing in the project's agent files points outside the repo or says "me".
  • For a project with UI, /dev/ui loads in development and its coverage test passes.
  • For an existing project, every project-specific rule from the old files still has a home.

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.