Better-Fullstack
AGENTS.md815 starsChanged 4 months agoMITView on GitHub
This file provides the small set of project-wide instructions agents should read before working in Better-Fullstack. Keep detailed, area-specific guidance in docs/guidelines/ instead of growing this file with task notes. For documentation work or when the right source of project knowledge is unclear, read docs/README.md. It defines document authority and routing. Do not bulk-read docs/guidelines/. Open a guideline when the request matches its topic. One exception is mandatory, not optional: adding any new library, tool, or category to any ecosystem…
We split this AGENTS.md into the rules it gives an agent
6 rules. About 0 are common, 2 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
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
- Very specific: line 8
Do not bulk-read docs/guidelines/.
- Distinctive: line 33
Prefer concise, simple solutions.
- Distinctive: line 34
If asked to do too much work at once, stop and say so clearly.
- Very specific: line 35
Use testing/README.md for prompt-free scaffolds, published-package validation, and runtime-proof lanes.
- Very specific: line 37
Never start the dev server (turbo dev, bun run dev, vite dev, etc.) unless explicitly asked to.
- Very specific: line 38
After code changes, run the smallest verification set that proves the modified area still works.
≈ “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.
Files
This file provides the small set of project-wide instructions agents should read before working in Better-Fullstack. Keep detailed, area-specific guidance in docs/guidelines/ instead of growing this file with task notes.
For documentation work or when the right source of project knowledge is unclear, read
docs/README.md. It defines document authority and routing.
Guidelines
Do not bulk-read docs/guidelines/. Open a guideline when the request matches its topic. One exception is mandatory, not optional: adding any new library, tool, or category to any ecosystem requires reading docs/guidelines/adding-new-tool-options/ first.
See docs/guidelines/ for deeper reference on these topics:
README.md- folder purpose, usage rules, and quick index for all guideline filesarchitecture-and-ownership.md- monorepo package ownership and where changes belongstack-graph-and-config-authority.md- graph authority, Role Binding ownership, flat projections, and graph-safe mutationslifecycle-commands-and-mcp.md- create/add/update/check semantics, plan/apply safety, baselines, and CLI/MCP paritygenerated-artifacts-and-sync.md- generated-file producers, authoritative inputs, and safe regeneration in dirty worktreestelemetry-privacy-and-internal-tools.md- telemetry allowlists, ingest privacy, aggregate access, and fail-closed internal toolspublic-docs-i18n-and-seo.md- public content authority, translation fallback, schema-derived references, and agent/SEO surfacescapability-evidence-levels.md- shared public proof levels and fail-closed claim rulesstack-options-and-compatibility.md- schema source of truth, canonical option metadata, aliases, and compatibility rulesgenerator-change-playbook.md- generated-output constraints, template changes, and verificationweb-builder-and-url-state.md- stack builder state, URL parsing, lazy-route constraints, and preview wiringtesting-release-and-upstream.md- targeted verification commands, release guard expectations, and upstream backport workflowpreview-publishing-security.md- PR preview threat model, trusted artifact boundary, environment setup, and workflow guardadding-new-tool-options/- read this subfolder when adding any new library, tool, or category to any ecosystem. Use its README for the source checklist, graph/CLI/MCP wiring, routing constraints, and verification
Web UI
apps/web has established primitives - tabs, copy buttons, command/code bars, cards. Before adding any UI control, grep apps/web/src/components for an existing one and reuse it. Do not introduce a second visual language for a control that already exists (no pills where the app uses tabs). New visual patterns need Ibrahim's sign-off.
Workflow
- Prefer concise, simple solutions. If a problem can be solved in a simpler way, propose it.
- If asked to do too much work at once, stop and say so clearly.
- Use
testing/README.mdfor prompt-free scaffolds, published-package validation, and runtime-proof lanes. - Never start the dev server (
turbo dev,bun run dev,vite dev, etc.) unless explicitly asked to. - After code changes, run the smallest verification set that proves the modified area still works. Prefer package-local
bun run lint,bun run test,bun run build, or specificbun test <file>commands over broad workspace sweeps. - For release-sensitive stack or generator changes, run
bun run testandbun run test:releasefrom the repo root.bun run testcovers template snapshots and CLI/builder parity;test:releasecovers the release validators.
Bun
Bun is the default package manager and script runner. Use bun install, bun run <script>, bun test, and bunx. Do not switch to npm, pnpm, yarn, npx, or ad hoc node wrappers unless a file explicitly requires it.
- In this WSL setup,
bunonPATHcan resolve to the Windows install instead of native Linux Bun. For Turbo runs and published-package verification, prefer~/.bun/bin/bunand~/.bun/bin/bunxexplicitly.
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 repository5 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- llms.txtllms.txt
- better-fullstackSkill
- frontend-designSkill
- web-design-guidelinesSkill
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.

