project-safety
Skill150 starsChanged 10 days agoMITView on GitHub
Install or reconcile research-project safety rules and protected folder boundaries for data, code, outputs, and paper sources. Use when a project lacks guardrails or is being migrated into the governed structure. Not for auditing the full research scaffold; use $audit-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-lightSkill
- 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
- 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
7 rules. About 0 are common, 7 are distinctive, and 0 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.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 9
CRITICAL RULE: Never delete data or code files. Never: Use legacy/ folder for originals.
- Distinctive: line 20
Before asking Claude to reorganize files
- Distinctive: line 21
When Claude will be running code or modifying data
- Distinctive: line 28
Add this to any project's CLAUDE.md file:
- Distinctive: line 76
Before Claude executes potentially destructive operations, ask for a preview:
- Distinctive: line 92
If correct: "Go ahead"
- Distinctive: line 93
If wrong: "Wait — don't do X, instead Y"
≈ “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
Project Safety Skill
CRITICAL RULE: Never delete data or code files. Never. Use legacy/ folder for originals. Copy, don't move.
Establish safety rules and structures before Claude makes changes to research projects.
Purpose
Based on Scott Cunningham's workflow: prevent accidental data loss by establishing rules and using legacy folders before Claude reorganizes or modifies project files.
When to Use
- Starting a new research project folder
- Before asking Claude to reorganize files
- When Claude will be running code or modifying data
- Setting up a project for collaborative work
Safety Rules Template
Add this to any project's CLAUDE.md file:
## Safety Rules
1. **Never delete data files** — No .csv, .dta, .xlsx, .parquet, or any data format
2. **Never delete code files** — No .py, .R, .do, .tex, .md or scripts
3. **Use legacy/ folder** — If reorganizing, move originals to legacy/ first
4. **Copy from legacy, don't move** — Always preserve the original
### If you need to reorganize:
1. Create a legacy/ folder if it doesn't exist
2. Move ALL original files into legacy/
3. Copy (not move) needed files into new structure
4. Never modify anything in legacy/Directory Structure
After safety setup:
project/
├── CLAUDE.md ← Safety rules + project context
├── README.md ← Project documentation
├── legacy/ ← PROTECTED: original files
│ └── [all originals]
├── code/
│ ├── R/
│ ├── python/
│ └── stata/
├── data/
│ ├── raw/ ← Copied from legacy, never modified
│ └── processed/
├── output/
│ ├── figures/
│ └── tables/
├── docs/
│ └── manuscript/
└── log/ ← Progress logsDry Run Pattern
Before Claude executes potentially destructive operations, ask for a preview:
"Tell me what commands you would run to reorganize this folder, but don't execute them yet."
"Show me what files would be affected by this change before you make it."
"Walk me through your plan for cleaning this data before you run any code."
When to use dry runs:
- File operations (move, delete, rename)
- Git operations (reset, clean, force push)
- Database operations
- Batch processing
- Any operation affecting multiple files
After reviewing:
- If correct: "Go ahead"
- If wrong: "Wait — don't do X, instead Y"
Workflow
- Create CLAUDE.md with safety rules
- Create legacy/ folder
- Move all originals to legacy/
- Copy needed files into new structure
- Verify legacy/ contains everything
- Proceed with project work
Prompt Template
I'm starting work on [PROJECT]. Before we do anything:
1. Create a CLAUDE.md with safety rules (never delete data/code, use legacy folder)
2. Create a legacy/ folder
3. Move all existing files into legacy/
4. Show me the proposed new directory structure before creating it
Do a dry run first — tell me what you would do before doing it.Why This Matters
- Claude operates at speed — mistakes happen fast
- You can't always verify what Claude will do before it does it
- Version control (git, Dropbox) helps but prevention is better
- The legacy/ folder is your safety net
- Dry runs give you a chance to catch mistakes
Example Use
"Set up my new Carbon Markets project with proper safety rules. Create the CLAUDE.md, legacy folder, and recommended directory structure. Show me your plan before executing."
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.

