longranger2/project-doc-bootstrap/project-doc-bootstrap
Claude Code and agent skill by longranger2/project-doc-bootstrap
- community
- ext:[{"type":"hook","description":"Pre/post lifecycle hooks"}]
Checking that the SKILL.md file is still in the repository.
Bootstrap a reusable repository documentation hierarchy for AI agents. Use when starting a new project, initializing repo docs, setting up CLAUDE.md and AGENTS.md, defining L1/L2/L3 documentation layers, creating AGENTS.md symlinks, or designing a fractal documentation system that helps AI quickly understand and safely update the codebase.
Install
npx vskill@latest install longranger2/project-doc-bootstrap/project-doc-bootstrapSemi-extensibleMentions customization, but needs skill-specific knowledge. Read the extensible skills standard
SKILL.md
Project Doc Bootstrap
Use this skill to set up a repository so future AI sessions can reach useful context in a few reads instead of re-deriving structure from source every time.
Default convention:
CLAUDE.mdis the source of truth.AGENTS.mdis a sibling symlink toCLAUDE.md.- Every directory that has a `CLAUDE.md` must also have an `AGENTS.md` symlink as a sibling. This is a hard invariant — never create one without the other.
- Docs follow a fractal hierarchy: L1 root guide, L2 directory guides, L3 file-header notes only where needed.
If the user explicitly wants different filenames or a different source-of-truth file, adapt the convention. Otherwise default to CLAUDE.md + AGENTS.md.
When to Use
Use this skill when the user wants to:
- initialize a new project's AI-facing docs
- set up
CLAUDE.md/AGENTS.md - design a directory-level documentation hierarchy
- make a repo easier for AI to understand and maintain
- standardize how docs are updated when files move or modules grow
- install a pre-commit hook that enforces doc-sync at commit time
Quick Start Workflow
- Read
references/blueprint.mdto choose the documentation model. - Read
references/workflow.mdto bootstrap the repo in the right order. - Read
references/templates.mdfor ready-to-copy templates: CLAUDE.md skeletons, symlink commands, and pre-commit hook scripts.
Core Outcomes
By the end of the task, the repo should have:
- one root
CLAUDE.mdthat explains architecture, rules, and the doc system - one root
AGENTS.md -> CLAUDE.mdsymlink - L2 docs for top-level and core module directories
- L2 docs for subdirectories that cross the documentation threshold
- explicit rules for when new L2 docs should be created later
- a pre-commit hook (
.git/hooks/pre-commit) that warns when staged source files are in a documented directory whoseCLAUDE.mdwas not also staged (hook is local-only, not committed) - a verification pass confirming symlink integrity, hook installation, and obvious gaps
Operating Rules
- Prefer small, stable documentation rules over exhaustive directory-by-directory prose.
- Keep each
SKILL.mdand each L2 doc concise; move detail into references when it grows. - Prefer symmetry for repeated trees such as locales, themes, provider categories, or parallel route groups.
- Do not create L2 docs for every leaf by default; stop when parent docs keep the tree understandable.
- Treat doc maintenance as part of code maintenance: if a directory changes shape, update its
CLAUDE.mdin the same task.
Decision Rules
Create an L2 doc when a directory meets any of these:
- it is a top-level or core module directory
- it has 4 or more direct child items or source files
- it is a clear module boundary or high-frequency entrypoint
- its parent doc can no longer explain it within 3 reads
Do not create an L2 doc when all of these are true:
- the directory has only 1-3 files
- it is a leaf route or narrow implementation detail
- its parent doc already provides enough context
Delivery Checklist
- [ ] Root
CLAUDE.mdexists and defines the doc hierarchy. - [ ] Root
AGENTS.mdis a symlink toCLAUDE.md. - [ ] Top-level/core directories have L2 docs where appropriate.
- [ ] Every directory with a `CLAUDE.md` has a sibling `AGENTS.md` symlink. This is a hard invariant — never create one without the other.
- [ ] Repeated structures are symmetric where that helps maintenance.
- [ ] New
AGENTS.mdfiles are symlinks, not duplicated files. - [ ]
.git/hooks/pre-commitexists and is executable (local-only, not committed). - [ ] The final audit lists any intentionally undocumented directories.
Final Verification
Before closing the task, run these checks in order and do not skip any:
- List every
CLAUDE.mdin the repo. - For each
CLAUDE.md, verify a siblingAGENTS.mdsymlink exists and points to thatCLAUDE.md. - Verify
.git/hooks/pre-commitexists and is executable. - Spot-check 2-3 representative L2 docs for format consistency.
- Report what was intentionally left undocumented and why.
Use the audit commands in references/templates.md to automate steps 1-2.
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingSKILL.md
Popularity
- Repository stars
- 2
- 7-day trend score
- 5
Stars and forks are for the whole GitHub repository, not this skill alone. Refreshed 2 Jun 2026, 5 months ago.
Extension points
hookPre/post lifecycle hooks
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/longranger2/project-doc-bootstrap/project-doc-bootstrap)