repro-lens
AGENTS.md44 starsChanged 47 days agoMITView on GitHub
Keep one implementation of each rule in src/repro_lens; CLI, hooks and skills use it. The checker must not import project code, execute notebook cells or start training. Keep review findings separate from confirmed policy violations. Do not call a clean static scan proof of reproducibility, or a two-run match proof of scientific validity. Add positive and negative behavioral cases when changing a detection rule. Preserve aliases, shadowed identifiers, configured exclusions and justified suppressions. Never autofix a scientific choice by inserting…
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,700 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repositoryOne other file it gives its agents.
- reproducibilitySkill
The file
We split this AGENTS.md into the rules it gives an agent
7 rules. About 0 are common, 4 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 3
Keep one implementation of each rule in src/repro_lens.
- Very specific: line 5
Keep review findings separate from confirmed policy violations.
- Distinctive: line 8
Add positive and negative behavioral cases when changing a detection rule.
- Distinctive: line 10
Never autofix a scientific choice by inserting seed 42 or relaxing a tolerance.
- Distinctive: line 13
Keep generated data/models and local evidence out of source control except tiny deliberately versioned fixtures.
- Very specific: line 16
Use uv run pytest, uv run ruff check .
- Distinctive: line 20
Keep Repro Lens free and community-supported.
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md 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
Working on Repro Lens
Keep one implementation of each rule in src/repro_lens; CLI, hooks and skills use it.
The checker must not import project code, execute notebook cells or start training.
Keep review findings separate from confirmed policy violations. Do not call a clean
static scan proof of reproducibility, or a two-run match proof of scientific validity.
Add positive and negative behavioral cases when changing a detection rule. Preserve aliases, shadowed identifiers, configured exclusions and justified suppressions. Never autofix a scientific choice by inserting seed 42 or relaxing a tolerance.
The template is a working user project: validate a generated copy, not just its text. Keep generated data/models and local evidence out of source control except tiny deliberately versioned fixtures. Document new limits alongside new capabilities.
Use uv run pytest, uv run ruff check . and uv run ruff format --check ..
Community support
Keep Repro Lens free and community-supported. Do not add paid pilots, consulting offers, pricing, donation links, GitHub Sponsors configuration or payment forms. The maintainer does not want to collect money through this project.
Product presentation
The maintainer expects a polished visual design and a useful first experience. Show a concrete problem, understandable evidence and a clear way to try the tool. Preserve readable code, keyboard access, mobile layouts and shareable example links. Review actual desktop and mobile rendering when changing the site. Prefer focused, working interactions over decorative controls or a long undifferentiated feature list. Claims of SOTA, superiority or adoption need comparative evidence; presentation alone does not establish them. Keep demo outcomes and live execution clearly distinguished.
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.