Claude-skills-kit project-onboarding

Full project onboarding for Claude Cowork: generates context.md, folder-instructions.md, file map, and starter prompts in one session. Triggers RU: «создай проект», «onboarding проекта», «новый проект», «настрой проект», «создай контекст проекта», «быстрый контекст», «обнови карту файлов», «пересканируй файлы». Triggers EN: 'create project', 'project onboarding', 'new project', 'set up project', 'quick context', 'update file map', 'rescan files'.

install

source · Clone the upstream repo

git clone https://github.com/KirKruglov/claude-skills-kit

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/KirKruglov/claude-skills-kit "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/project-onboarding" ~/.claude/skills/kirkruglov-claude-skills-kit-project-onboarding && rm -rf "$T"

manifest: skills/project-onboarding/SKILL.md

source content

Skill: project-onboarding

Complete project onboarding for Claude Cowork in a single session.

Triggers

Russian: «создай проект», «onboarding проекта», «новый проект», «настрой проект», «создай контекст проекта», «быстрый контекст», «обнови карту файлов», «пересканируй файлы» English: "create project", "project onboarding", "new project", "set up project", "create project context", "quick context", "update file map", "rescan files"

Language Detection

Determine the language of the user's request:

If the request is in Russian → use templates with
```
-ru
```
suffix
Otherwise → use templates with
```
-en
```
suffix

All output (headings, labels, comments, instructions) must match the detected language.

Modes

Mode	Triggers	Actions
`new`	"create project", "onboarding", "new project", "set up project"	Full interview → context.md + folder-instructions.md + file map + prompts
`quick`	"quick context", "create project context"	Blocks 1–3 + auto-scan → context.md
`scan`	"update file map", "rescan files"	File Map module only → update section in context.md

Instructions

Step 1 — Determine Mode

If trigger matches

scan

→ go directly to the File Map module. If

quick

→ interview blocks 1–3, then File Map. Otherwise →

new

, all steps in sequence.

Step 2 — Interview (modes: new / quick)

Ask questions in blocks. Wait for a response after each block. Keep questions concise.

Block 1 — Basics

1. Project name?
2. Type: analytics / content / business / personal / other?
3. Status: Active / On hold / Completed?
4. Start date? Deadline?

Block 2 — Project Overview

5. Describe the project in 2–4 sentences.
6. Vision — what outcome = success?

Block 3 — Current Phase

7. What are we working on right now?
8. What is the concrete deliverable for this phase?

→ Mode

quick

stops here. Proceed to the File Map module.

Block 4 — Structure

9. Key elements / components / layers of the project?

Block 5 — Stakeholders

10. Stakeholders, partners, external dependencies?

Block 6 — Constraints

11. Constraints: time, resources, risks?

Block 7 — Open Questions

12. Known unresolved tasks or questions?

Block 8 — Additional (optional)

13. Additional sections needed: market context / methodology / roadmap?

If yes — ask 1–2 clarifying questions for each selected section.

Step 3 — File Map Module

Determine the project folder:

If a user folder is mounted (mounted workspace) — use its root.
If a path was mentioned during the interview — use it.
Otherwise — ask: "Provide the path to the project folder."

Run:

find [project-folder] -type f \
  -not -path '*/.git/*' \
  -not -path '*/node_modules/*' \
  -not -path '*/__pycache__/*' \
  -not -name '.DS_Store' \
  | head -200

For each file determine: path, extension, size. For text files (.md, .txt, .csv) — read the first line as the description. For all others — use

[auto]

based on filename or

[clarify]

Group by folder. Format the section as:

# Data Structure

## [folder-name]/
| File | Type | Size | Description |
|------|------|------|-------------|
| filename.ext | EXT | X KB | description |

scan

mode — locate the existing

context.md

and replace the "Data Structure" section. Do not modify anything else.

Step 4 — Generate context.md (modes: new / quick)

Rules:

Do not copy answers verbatim — write coherent structured text
Preserve the user's terminology
Do not infer — use only facts from the answers
Missing information →
```
[clarify]
```
Use the File Map module output for the "Data Structure" section

Template:

resources/context-template-ru.md

(Russian) or

resources/context-template-en.md

(English)

Output:

context.md

Step 5 — Rules Module (mode: new only)

Load the template from

resources/rules-templates-ru.md

resources/rules-templates-en.md

based on the detected language and project type (answer 2). Add general rules (file discovery, language, behavior). Adapt to the user's answers.

Output:

folder-instructions.md

After generating, ask: "Rules are ready. Would you like to edit or add your own?"

Step 6 — Starter Prompts Module (mode: new only)

Ask: "Generate starter prompt templates for this project type?" If yes — load templates from

resources/prompt-templates-ru.md

resources/prompt-templates-en.md

based on the detected language and project type. Adapt to the project context (project name, terminology).

Output: files in

resources/prompts/

Step 7 — Summary

Show the table:

| File | Status |
|------|--------|
| context.md | Created, X blocks filled, Y [clarify] |
| folder-instructions.md | Created / Skipped |
| resources/prompts/*.md | N templates created / Skipped |

Provide all files for download.

Constraints

Maximum 200 files in auto-scan (if more → warn and show top-level only)
This skill does not generate prompt-builder prompts — that is a separate skill
Do not modify project files other than context.md (in scan mode)