init-project-light
Skill150 starsChanged 9 days agoMITView on GitHub
Bootstrap a lightweight project with minimal guidance, context, and repository structure. Use when a small non-research project needs durable AI collaboration without the full research or course scaffold. Not for formal research projects; use $init-project-research.
In this repository89 other files it gives its agents.
- AGENTS.mdAGENTS.md
- docs/agents.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- beamer-deckSkill
- bib-coverageSkill
- bib-filterSkill
- bib-parseSkill
- brief-compliance-checkSkill
- camera-readySkill
- causal-designSkill
- checkpointSkill
- code-archaeologySkill
- compile-knowledgeSkill
- computational-experimentsSkill
- cross-language-checkSkill
- data-analysisSkill
- devils-advocateSkill
- docs-consistencySkill
- docxSkill
- experiment-designSkill
- grill-meSkill
- handoffSkill
- ideasSkill
- init-project-courseSkill
- init-project-orchestrationSkill
- init-projectSkill
- insights-deckSkill
- interview-meSkill
- knowledge-lintSkill
- latex-diffSkill
- latex-health-checkSkill
- latex-polishSkill
- latex-postersSkill
- latex-scaffoldSkill
- latexSkill
- latex-templateSkill
- lean-checkSkill
- math-proofSkill
- mcp-builderSkill
- method-auditSkill
- multi-perspectiveSkill
- numerical-checkSkill
- pdfSkill
- pipeline-manifestSkill
- playwright-cliSkill
- postmortemSkill
- pre-commit-auditSkill
- pre-submission-reportSkill
- project-deckSkill
- project-safetySkill
- proof-readabilitySkill
- proofreadSkill
- python-envSkill
- quarto-deckSkill
- reorgSkill
- replication-auditSkill
- replication-packageSkill
- retarget-journalSkill
- review-clusterSkill
- review-responseSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
11 rules. About 0 are common, 7 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
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
- Distinctive: line 17
When the user says "set up something light", "quick init", "organise this folder"
- Distinctive: line 30
Read everything already in the directory before asking questions.
- Very specific: line 33
Read text files (.md, .tex, .bib, .txt) to understand content — respect file size (skip files > 500 lines, note them)
- Very specific: line 42
Use the available structured-question mechanism.
- Distinctive: line 50
If Phase 1 gave you enough, confirm your understanding instead of asking:
- Very specific: line 57
Follow the lean-guidance-files rule.
- Distinctive: line 86
Never create more than 2-3 folders — this is a light project
- Distinctive: line 87
Never move files without explicit approval
- Distinctive: line 88
If the existing structure already makes sense, say so and skip this phase
- Very specific: line 89
If there's a .claude/settings.local.json, leave it.
- Distinctive: line 112
Only create if missing. Never overwrite existing permissions.
≈ “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.
Files
Init Project Light
Lightweight project bootstrapper for small projects that do not need a full research-project scaffold.
When to Use
- Small document collections (proposals, applications, meeting notes)
- One-off or short-lived projects
- Projects without a code pipeline or Overleaf link
- When the user says "set up something light", "quick init", "organise this folder"
- Any project that does not warrant the installation's full research-project initializer
When NOT to Use — Escalate to a full research-project initializer
- Research papers targeting a journal or conference
- Projects with code, data, or computational pipelines
- Anything that needs Overleaf, git, or a vault atlas entry
Phase 1: Scan
Read everything already in the directory before asking questions.
- List all files and folders (excluding
.claude/,.DS_Store) - Read text files (
.md,.tex,.bib,.txt) to understand content — respect file size (skip files > 500 lines, note them) - Build a mental model: what is this project, what's the main output, who's involved?
Goal: Minimise interview questions by inferring answers from existing files.
Phase 2: Interview (2-3 questions max)
Use the available structured-question mechanism. Only ask what you couldn't infer from Phase 1.
Pick from these (skip any you can already answer):
- What is this project? — one sentence (e.g., "PhD research proposal for [University]")
- What's the main output? — document, application, collection of notes, etc.
- Anyone else involved? — names and roles if relevant
If Phase 1 gave you enough, confirm your understanding instead of asking:
"From the files, this looks like [X]. The main output is [Y]. Correct?"
Phase 3: Create CLAUDE.md
Follow the lean-guidance-files rule. Include only:
- Project overview — 2-3 sentences from interview/scan
- People — if collaborators/supervisors exist
- Directory structure — compact tree of what exists
- Conventions — only if detectable (e.g., LaTeX compilation, bibliography style)
- Key context — anything a future session needs to know immediately
Do NOT include:
- Detailed literature notes or reference lists
- Action items or timelines (those go to vault)
- Anything that duplicates global rules
Phase 4: Organise (suggest, don't force)
Based on what's in the directory, suggest lightweight organisation. Present options and wait for approval.
Standard suggestions
| Folder | When to suggest |
to-sort/ | Multiple unsorted documents exist |
docs/ | Reference materials, guidelines, or background reading present |
archive/ | Old versions or abandoned drafts detected |
Rules
- Never create more than 2-3 folders — this is a light project
- Never move files without explicit approval
- If the existing structure already makes sense, say so and skip this phase
- If there's a
.claude/settings.local.json, leave it. If not, create one with standard permissions.
Standard permissions (.claude/settings.local.json)
{
"permissions": {
"allow": [
"Bash(latexmk *)",
"Bash(ls:*)",
"Bash(mkdir:*)",
"Bash(tree:*)",
"Edit",
"Glob",
"Grep",
"Read",
"Write"
],
"deny": []
}
}Only create if missing. Never overwrite existing permissions.
Phase 5: Confirmation
Short report:
Set up lightweight project: <name>
Created:
- CLAUDE.md
- [any folders created]
- [.claude/settings.local.json if created]
Skipped (use an installed full research-project initializer if needed later):
- Git, Overleaf, vault atlas, code scaffoldCross-References
| Skill | Relationship |
| Installed full research-project initializer | Escalate to this for full research projects |
update-project-doc | Run later to refresh CLAUDE.md if the project grows |
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
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.

