agentleFS
aka-kikaaka-kika/akakika-skills/skills/project-intelligence/explain-new-project/SKILL.md

explain-new-project

Skill11 starsChanged 3 months agoMITView on GitHub

Use when the user adds a project to the workspace, opens an unfamiliar repo, asks "what is this project", "explain this codebase", or invokes explain-new-project. Explores a newly opened workspace project and writes a plain-language report for a self-taught, non-technical user—quick start (what it is + how to run) first, then optional depth—and always saves PROJECT_GUIDE.md in the project root.

  • Reads credentials

The file

Show its rules and text

We split this skill into the rules it gives an agent

19 rules. About 0 are common, 15 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈15

    Distinctive · estimate

    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
  1. Distinctive:

    Write the report in the same language the user uses in chat (default: English).

    line 15
  2. Distinctive:

    No assumed CS degree

    line 19
  3. Distinctive:

    Write the full report using the template below.

    line 27
  4. Distinctive:

    Use search/read tools; prefer evidence from files over assumptions.

    line 45
  5. Distinctive:

    Inline teach: add a Learn: line immediately after, max two sentences.

    line 51
  6. Distinctive:

    Do not dump a glossary at the end unless there are many terms.

    line 53
  7. Distinctive:

    Prefer "this folder holds…" over "the architecture implements…"

    line 107
  8. Distinctive:

    If something is incomplete or broken, say so plainly

    line 108
  9. Distinctive:

    Do not lecture about best practices unless it affects *their* next step

    line 109
  10. Distinctive:

    Do not output raw directory trees without explanation

    line 113
  11. Distinctive:

    Do not recommend large refactors or "you should learn X framework"

    line 114
  12. Distinctive:

    Do not assume they know git beyond basics unless the project requires it

    line 115
  13. Very specific:

    Do not skip saving PROJECT_GUIDE.md

    line 116
  14. Distinctive:

    Do not ask whether to save the file

    line 117
  15. Distinctive:

    If several projects are open:

    line 121
  16. Very specific:

    If the user named one, explain only that root and save PROJECT_GUIDE.md there.

    line 122
  17. Distinctive:

    If unclear, list each root in one line, ask which to explain, or give a one-minute summary per root and save one guide per root they care …

    line 123
  18. Very specific:

    Run npm install once (downloads dependencies).

    line 131
  19. Very specific:

    Run npm run dev and open the link it prints.

    line 132

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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.

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

Explain New Project

Audience

The reader is self-taught (about two years of hands-on work), not tech-savvy, and learns best from everyday language. They are capable and curious—not beginners who need hand-holding, but people who want clarity without jargon walls.

Language rule

Write the report in the same language the user uses in chat (default: English). If they mix languages, follow their latest message.

"My own language" means:

  • Short sentences, active voice
  • No assumed CS degree
  • Technical terms only when needed—and always explained (see Jargon rule)
  • Analogies welcome when they clarify (one line max)

When invoked

  1. Confirm which folder is the project root (workspace root, or a path the user names).
  2. Explore before writing—do not guess. Read enough to be accurate.
  3. Write the full report using the template below.
  4. Always save PROJECT_GUIDE.md at the project root (same folder as README or main manifest). Overwrite if it already exists.
  5. In chat: show the Quick start sections in full, then one line pointing to the file for the rest—e.g. Full guide saved to PROJECT_GUIDE.md—scroll down there for the map, tech stack, gotchas, and next steps.

Exploration checklist

Work through these until you have a confident picture (skip what does not exist):

- [ ] README, README.*, docs/, AGENTS.md, CONTRIBUTING
- [ ] Top-level folders (what lives where)
- [ ] Manifest: package.json, pyproject.toml, Cargo.toml, go.mod, Podfile, etc.
- [ ] Entry points: main app file, index, App.tsx, src/main.*
- [ ] How to run: scripts in package.json, Makefile, docker-compose, .env.example
- [ ] Tests folder (if any)—only to say *whether* tests exist, not to teach testing
- [ ] .gitignore / deploy hints (Vercel, Docker, etc.)—plain words only

Use search/read tools; prefer evidence from files over assumptions.

Jargon rule

Whenever you use a term the reader might not know (API, dependency, build, deploy, framework, repo, branch, env var, etc.):

Inline teach: add a Learn: line immediately after, max two sentences.

Do not dump a glossary at the end unless there are many terms; teach inline as you go.

Report template

Order matters. Quick start first so the reader can stop after section 2 and still be unblocked.

# [Project name] — plain guide

> **Quick start** — read through "How you run it", then stop if that is enough.
> **The rest** — optional depth when you want the full picture.

---

## In one minute
[2–4 sentences: what this project is for, who it is for, and what "done" looks like]

## How you run it
[Only steps that exist in the repo. Numbered steps. If setup is unclear, say what is missing.]

**Learn:** [only if "run", "install", or "dev server" needs explaining]

---

## The rest (optional)

### What you would use this for
[Concrete scenarios in plain language]

### How it is organized (the map)
[Folder-by-folder in plain English—what each important folder *is for*, not every file]

### Main pieces (the cast)
| Piece | Plain role |
|-------|------------|
| ... | ... |

### What technologies it uses (no buzzwords)
[List stack in friendly terms, e.g. "website UI in React", "small server in Node"]

### Things that might confuse you
[2–5 honest gotchas: env files, monorepo, generated folders, old README, etc.]

### Safe next steps for you
[3–5 ordered suggestions: e.g. open X file, run Y command—matched to their level]

### If you want to go deeper later
[1–3 file paths worth opening when curious—not a reading list homework]

Tone

  • Respectful, direct, warm—not childish
  • Say "you" to the reader
  • Prefer "this folder holds…" over "the architecture implements…"
  • If something is incomplete or broken, say so plainly
  • Do not lecture about best practices unless it affects their next step

What not to do

  • Do not output raw directory trees without explanation
  • Do not recommend large refactors or "you should learn X framework"
  • Do not assume they know git beyond basics unless the project requires it
  • Do not skip saving PROJECT_GUIDE.md
  • Do not ask whether to save the file

Multi-root workspaces

If several projects are open:

  • If the user named one, explain only that root and save PROJECT_GUIDE.md there.
  • If unclear, list each root in one line, ask which to explain, or give a one-minute summary per root and save one guide per root they care about.

Example (tone only)

In one minute: Shop Helper is a small web app that lets you track products and prices. When you open it in the browser, you see a list you can add to and edit.

How you run it:

  1. Open a terminal in this folder.
  2. Run npm install once (downloads dependencies).
  3. Run npm run dev and open the link it prints.

Learn: npm is a tool that installs and runs JavaScript project pieces; you use the same two commands in most web projects like this.

Follow-up skills

OrderSkillWhenSaves
Optionalproject-worth-my-timeIs this worth continuing?PROJECT_VERDICT.md
On returnproject-catch-upComing back after a break—updates, tools, skillsPROJECT_CATCH_UP.md

Invoke the project-worth-my-time or project-catch-up skill when you reach those points.


For more output samples, see examples.md.

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.