gravidex
AGENTS.md1 starsChanged 7 days agoMITView on GitHub
These rules are automatically loaded into every conversation turn. They eliminate the need for manual slash commands by enforcing core disciplines autonomously.
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 repository62 other files it gives its agents.
- .claude/skills/humanizer/AGENTS.mdAGENTS.md
- agentshieldSkill
- asd-ste100Skill
- autonomous-agent-engineeringSkill
- awwwardsSkill
- backend-engineeringSkill
- boarduiSkill
- design-systemsSkill
- frontend-engineeringSkill
- interface-designSkill
- jurySkill
- karpathy-guidelinesSkill
- last30daysSkill
- lemo-opuscarSkill
- motionsites-web-builderSkill
- quality-engineeringSkill
- rev-engineerSkill
- scraplingSkill
- security-auditSkill
- skillforgeSkill
- stageshowSkill
- technical-documentationSkill
- agentshieldSkill
- anti-vcodeSkill
- arenaSkill
- asd-ste100Skill
- autonomous-agent-engineeringSkill
- awwwardsSkill
- backend-engineeringSkill
- boarduiSkill
- courtSkill
- designsSkill
- design-systemsSkill
- diagram-designSkill
- frontend-engineeringSkill
- humanizerSkill
- impeccableSkill
- interface-designSkill
- jurySkill
- karpathy-guidelinesSkill
- last30daysSkill
- lemo-opuscarSkill
- logo-designSkill
- model-routerSkill
- motionsites-web-builderSkill
- ponytailSkill
- quality-engineeringSkill
- rev-engineerSkill
- rigorSkill
- sanzo-wada-colorsSkill
- scraplingSkill
- secure-auditSkill
- security-auditSkill
- skillforgeSkill
- socraticSkill
- stageshowSkill
- stealSkill
- tasteSkill
- technical-documentationSkill
- threeuiSkill
- The first 60 shown.
The file
We split this AGENTS.md into the rules it gives an agent
6 rules. About 0 are common, 4 are distinctive, and 2 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.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 9
Standard Library First: Never introduce third-party dependencies or pip/npm packages when native platform capabilities or Python/Node …
- Distinctive: line 10
Clean Scope: Do not reformat, rewrite, or touch unrelated functions, comments, or whitespace.
- Distinctive: line 15
Zero Hallucinated Completion: Never declare a task, bug fix, or refactor complete until you have executed real verification tests on disk.
- Very specific: line 16
Check Evidence: Run python3 -B test_all.py or the specific test suite.
- Very specific: line 31
Automatic 429 Cascade: If an external API returns HTTP 429 (Rate Limited) or quota exhaustion, automatically cascade to the next configured …
- Distinctive: line 36
Do not output emoji characters in code, documentation, test suites, terminal logs, or git commit messages.
≈ “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
Autonomous Engineering Rules for Antigravity
These rules are automatically loaded into every conversation turn. They eliminate the need for manual slash commands by enforcing core disciplines autonomously.
1. Minimal Diff & YAGNI Principle (Ponytail Engine)
- Smallest Complete Diff: Choose the solution with the fewest changed lines, fewest files touched, and zero unnecessary abstractions.
- Standard Library First: Never introduce third-party dependencies or pip/npm packages when native platform capabilities or Python/Node standard libraries suffice.
- Clean Scope: Do not reformat, rewrite, or touch unrelated functions, comments, or whitespace.
2. Empirical Verification Gate (Rigor Engine)
- Zero Hallucinated Completion: Never declare a task, bug fix, or refactor complete until you have executed real verification tests on disk.
- Check Evidence: Run
python3 -B test_all.pyor the specific test suite. Inspect exit codes (must be 0) and verify that all test assertions passed. - Log Proof: Report the exact runtime and test count in your final summary.
3. Anti-Slop Frontend & Typographic Discipline (Anti-VCode)
- Banned Patterns: Forbid generic AI aesthetics: centered purple/cyan glow effects, gratuitous glassmorphism, floating cards without grid purpose, and generic placeholder copy.
- Design Foundations: Enforce Swiss grid discipline, restrained motion (< 300ms, standard easing), and accessible contrast (WCAG AAA/AA).
4. Multi-Model Resilience & Automatic Failover (Model Router)
- Task Specialization:
- Complex algorithmic edge cases, mathematical reasoning, or race conditions: route to
deepseek/deepseek-r1. - Fast boilerplate, bulk refactoring, and file transforms: route to
qwen/qwen-2.5-coder-32bor Groq.
- Complex algorithmic edge cases, mathematical reasoning, or race conditions: route to
- Automatic 429 Cascade: If an external API returns
HTTP 429(Rate Limited) or quota exhaustion, automatically cascade to the next configured fallback provider without halting workflow momentum.
5. Strict Emoji Ban
- Do not output emoji characters in code, documentation, test suites, terminal logs, or git commit messages.
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.