skillsboard
AGENTS.md10 starsChanged 3 months agoMITView on GitHub
- Reads credentials
- Commits and pushes
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 repository69 other files it gives its agents.
- public/llms.txtllms.txt
- ab-testingSkill
- ad-creativeSkill
- adsSkill
- agent-email-inboxSkill
- ai-seoSkill
- analyticsSkill
- animation-vocabularySkill
- apple-designSkill
- asoSkill
- attributionSkill
- churn-preventionSkill
- cold-emailSkill
- co-marketingSkill
- community-marketingSkill
- competitor-profilingSkill
- competitorsSkill
- content-strategySkill
- copy-editingSkill
- copywritingSkill
- croSkill
- customer-researchSkill
- design-taste-frontendSkill
- directory-submissionsSkill
- email-best-practicesSkill
- emailsSkill
- emil-design-engSkill
- find-animation-opportunitiesSkill
- free-toolsSkill
- grill-meSkill
- imageSkill
- improve-animationsSkill
- influencer-marketingSkill
- interactivity-best-practicesSkill
- launchSkill
- lead-magnetsSkill
- marketing-councilSkill
- marketing-ideasSkill
- marketing-loopsSkill
- marketing-planSkill
- marketing-psychologySkill
- next-cache-components-optimizerSkill
- offersSkill
- onboardingSkill
- paywallsSkill
- popupsSkill
- pricingSkill
- product-marketingSkill
- programmatic-seoSkill
- prospectingSkill
- public-relationsSkill
- react-emailSkill
- referralsSkill
- remocnSkill
- remotion-best-practicesSkill
- resend-cliSkill
- resendSkill
- review-animationsSkill
- revopsSkill
- sales-enablementSkill
- The first 60 shown.
The file
We split this AGENTS.md into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 13
After changing .env.local, restart pnpm dev so the pg pool in lib/db/index.ts (created at module load) picks up the new DATABASE_URL.
≈ “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
AGENTS.md
Cursor Cloud specific instructions
What this is
skillsboard is a single Next.js 16 (App Router, Turbopack) app — UI + API routes in one process. Its backing store is a Neon PostgreSQL database (pulled from Vercel; see below). Auth is Better Auth (email OTP via Resend + organizations + OAuth/MCP provider). There is no separate backend to run.
Running locally
- Environment lives in
.env.local(gitignored, so it is not in the repo). Populate it from Vercel (VERCEL_TOKENis provided as a secret): the project is linked totommasos-projects-bb9d6551/skillsboard. Pull the Development variables (Neon PostgresDATABASE_URL/DATABASE_URL_UNPOOLED,BETTER_AUTH_SECRET,VERCEL_OIDC_TOKEN, etc.) with:npx vercel env pull .env.local --environment=development --yes(runnpx vercel link --yes --project skillsboard --scope tommasos-projects-bb9d6551first if.vercel/project.jsonis absent). This points the app at the persistent Neondevelopmentbranch, isolated from Production'smainbranch. Preview deployments use Neon-managed ephemeral branches. SetBETTER_AUTH_URL=http://localhost:3000in.env.localfor local auth callbacks. PostHog variables are scoped to Vercel Production and analytics is intentionally disabled in local development and Preview deployments. - Dev server:
pnpm dev(serveshttp://localhost:3000). Standard scripts live inpackage.json. - After changing
.env.local, restartpnpm devso thepgpool inlib/db/index.ts(created at module load) picks up the newDATABASE_URL.
Database schema
The schema lives in lib/db/schema.ts and is managed with versioned Drizzle migrations committed in drizzle/ (see docs/database-migrations.md). After editing the schema, run pnpm db:generate --name <description> and commit the generated SQL together with the schema change — do not hand-edit drizzle/meta/. Vercel builds apply pending migrations automatically (vercel-build runs pnpm db:migrate before next build): preview builds against the Neon preview/<git-branch> database branch, production builds against Neon main. Drizzle uses the direct DATABASE_URL_UNPOOLED; the app continues to use pooled DATABASE_URL. To align the database in .env.local (normally the Neon development branch), run pnpm db:migrate. pnpm db:push remains for throwaway prototyping only; never push to a migration-managed database, or the next db:migrate will fail on drift.
Gotchas
pnpm lintrunseslint ., but ESLint is not a declared dependency and there is no ESLint config, so it fails out of the box (not a code problem). For type checking usenpx tsc --noEmit.drizzle-kit@0.31.10mis-introspects empty PostgreSQL array defaults. A repeatedpnpm db:push -- --strict --verbosemay propose only idempotentSET DEFAULT '{}'statements forskill.tagsandskill.examplePromptseven when both defaults are already correct. Abort if it proposes anything else.- The shadcn CLI is not a dependency. Run it pinned with
pnpm dlx shadcn@4.23.0 add <component>, also for the remocn registry wherever the remocn skill saysshadcn add @remocn/<name>. Theshadcn/tailwind.cssstylesheet the components rely on is vendored atapp/styles/shadcn-tailwind.css; when moving to a newer shadcn, copy that version'sdist/tailwind.cssover it and update the version in its header and in these commands. - Skill metadata is fetched live from the GitHub REST API. It works unauthenticated; set
GITHUB_TOKENto avoid rate limits. - In development Better Auth sets cookies with
sameSite: "none"; secure: true(seelib/auth.ts), which can affect session behavior on plainhttp://localhostin some browsers. - In development, email OTP skips Resend and accepts any 6-digit code after the “Continue” step (see
lib/auth.ts). Preview/production send real codes via Resend. - Pull requests: open them as drafts (
gh pr create --draft), link the demo video withpnpm pr:video <pr-number> <file>, thengh pr ready <pr-number>. The "Demo video" check skips drafts and runs onready_for_review; opening ready without a video leaves a red run. - Signing in on a Vercel preview for a demo video: request a sign-in code for
demo-video@skillsboard.invalid(POST /api/auth/email-otp/send-verification-otpwith{"email":"demo-video@skillsboard.invalid","type":"sign-in"}; no email is sent), thenPOST /api/auth/sign-in/email-otpwith the same email andotpset to the value of the Preview-onlyPREVIEW_TEST_SIGN_IN_SECRET. Send anOriginheader matching the preview URL. The guard isgetPreviewTestSignInSecretinlib/auth-environment.ts: it requiresVERCEL_ENV === "preview"exactly and a secret of at least 32 characters, so it is off in production, development and locally. Never add that variable to Production. - The Discover/catalog feature calls the external skills.sh API via Vercel OIDC; it degrades gracefully (shows "catalog unavailable") when unavailable locally and does not block the core flow.
<!-- BEGIN:nextjs-agent-rules -->This is NOT the Next.js you know
This version has breaking changes — APIs, conventions, and file structure may all differ from your training data. Read the relevant guide in node_modules/next/dist/docs/ (resolved from this file's directory; in monorepos the next package may not be visible from the repo root) before writing any code. Heed deprecation notices.
This block is written and re-added by next dev — verify at node_modules/next/dist/server/lib/generate-agent-files.js. Removing it from a diff only re-creates the uncommitted change; committing it with your work keeps the tree clean.
<!-- END:nextjs-agent-rules -->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.

