soumatheusgomessoumatheusgomes/vibe-coding-toolkit/AGENTS.md

vibe-coding-toolkit

AGENTS.md2k starsChanged 14 days agoMITView on GitHub

This repository distributes documentation, prompts, and templates. It is not a Next.js application; projects are generated in separate directories.

We split this AGENTS.md into the rules it gives an agent

4 rules. About 2 are common, 2 are distinctive, and 0 are very specific to this project.

≈2
Common · estimate
Many files like this have a rule of this kind.
≈2
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

Common:
Use current official sources and Context7 where relevant.
line 12
Distinctive:
Keep public paths in English.
line 18
Common:
Do not copy credentials, private domains, dumps, personal caches, or tightly coupled harnesses into public examples.
line 23
Distinctive:
Do not add Git blockers, branch/worktree requirements, or commit/push hooks.
line 24

≈ “Common” is an estimate: at least 1 in 5 of the 36,368 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.

Keep what fits and drop the rest. Your version, ready for your agents.

Files

AGENTS.md

Vibe Coding Toolkit

This repository distributes documentation, prompts, and templates. It is not a Next.js application; projects are generated in separate directories.

Scope and sources

Preserve the entry points: machine setup, project bootstrap, and tool-specific adoption.
Recipe: docs/04-project-bootstrap.md; defaults: docs/06-project-defaults.md; entry point: docs/prompts/11-next-project-bootstrap.md.
Catalog: docs/tools/16-vercel-skills.md. Use the name from SKILL.md, rather than inferring it from the directory name.
Generated documents: templates/project-docs/manifest.json, with four root files and five files in docs/. Product destination: PRODUCT.md; preserve the historical URL templates/product.md.template.
Portable instructions: templates/AGENTS.md.template; the Claude template imports that file.
Use current official sources and Context7 where relevant. Never send secrets in queries.
English is the canonical language for documentation, prompts, templates, examples, and assets. Only README.pt-BR.md and README.es.md are localized; keep them aligned with README.md. Agent conversations follow the user's language.

Changes and validation

Reuse templates and rules. Do not add an application, custom generator, or product dependencies at the root.
Keep public paths in English; update links, anchors, indexes, recipes, and examples together when their contracts change.
Validate executable templates in a temporary project; check local links, configuration syntax, and git diff --check.
Validator regression tests: node --test scripts/validate-docs.test.mjs. Links, anchors, manifest, and JSON: node scripts/validate-docs.mjs.
Reproducible baseline validation: node scripts/validate-next.mjs with Node.js 24, npm, and network access. It creates and removes the temporary project; dependencies stay outside the root.
Rule self-check: node templates/eslint/verify.mjs templates/eslint/eslint-rules/index.cjs, in an environment with ESLint available.
Do not copy credentials, private domains, dumps, personal caches, or tightly coupled harnesses into public examples.
Do not add Git blockers, branch/worktree requirements, or commit/push hooks.
Publishing and pushing require an explicit request. The bootstrap does not publish the application.

Agents

Ponytail full: the smallest complete change, preserving security, accessibility, and relevant checks.
Caveman: concise conversation in the user's language; normal wording for code, documentation, and commits.
Respect existing authorization and directly requested execution; do not repeat design/plan approvals.
Public content works with different LLMs. Configure Claude and Codex separately; do not hardcode a personal model, provider, or authentication method in templates.
Preserve the session's client, model, and authentication; do not enable proxies or API billing as a side effect of Graphify/Caveman.
Relevant, supported delegation specifies the authorized model, files, and completion criteria.

Download now (ZIP)

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 28,693 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.
- CLAUDE.mdCLAUDE.md

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

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.

Check your setup, free