Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Skill from get-convex/agent-skills
.claude/skills/convex-setup-auth/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-17 | ✗→✓ | ▲ Improved | — | — |
| case-16 | ✗→✓ | ▲ Improved | — | — |
| case-08 | ✗→✓ | ▲ Improved | — | — |
| case-04 | ✗→✓ | ▲ Improved | — | — |
| case-01 | ✗→✓ | ▲ Improved | — | — |
Implement secure authentication in Convex with user management and access control.
JWT)
one-line fix
Convex supports multiple authentication approaches. Do not assume a provider.
Before writing setup code:
makes it obvious
user wants to switch
ask before proceeding
Common options:
the user wants auth handled directly in Convex
Clerk or the user wants Clerk's hosted auth features
already uses WorkOS or the user wants AuthKit specifically
Auth0
above
Look for signals in the repo before asking:
@clerk/*, @workos-inc/*, @auth0/*, or Convex Authpackages
convex/auth.config.ts, auth middleware, providerwrappers, or login components
Read the provider's official guide and the matching local reference file:
references/convex-auth.md
references/clerk.md
references/workos-authkit.md
references/auth0.md
The local reference files contain the concrete workflow, expected files and env vars, gotchas, and validation checks.
Use those sources for:
convex/auth.config.ts setupFor shared auth behavior, use the official Convex docs as the source of truth:
ctx.auth.getUserIdentity()
for optional app-level user storage
authorization guidance
provider is Convex Auth
Prefer official docs over recalled steps, because provider CLIs and Convex Auth internals change between versions. Inventing setup from memory risks outdated patterns. For third-party providers, only add app-level user storage if the app actually needs user documents in Convex. Not every app needs a users table. For Convex Auth, follow the Convex Auth docs and built-in auth tables rather than adding a parallel users table plus storeUser flow, because Convex Auth already manages user records internally. After running provider initialization commands, verify generated files and complete the post-init wiring steps the provider reference calls out. Initialization commands rarely finish the entire integration.
The most common auth task is checking identity in Convex functions.
ts// Bad: trusting a client-provided userId export const getMyProfile = query({ args: { userId: v.id("users") }, handler: async (ctx, args) => { return await ctx.db.get(args.userId); }, });
ts// Good: verifying identity server-side export const getMyProfile = query({ args: {}, handler: async (ctx) => { const identity = await ctx.auth.getUserIdentity(); if (!identity) throw new Error("Not authenticated"); return await ctx.db .query("users") .withIndex("by_tokenIdentifier", (q) => q.eq("tokenIdentifier", identity.tokenIdentifier), ) .unique(); }, });
storage, and authorization patterns
app needs them
configuration if requested
If the flow blocks on interactive provider or deployment setup, ask the user explicitly for the exact human step needed, then continue after they complete it. For UI-facing auth flows, offer to validate the real sign-up or sign-in flow after setup is done. If the environment has browser automation tools, you can use them. If it does not, give the user a short manual validation checklist instead.
references/convex-auth.mdreferences/clerk.mdreferences/workos-authkit.mdreferences/auth0.mdpatterns
users table or storeUser flow forConvex Auth
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-14 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-17 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-21 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-22 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-16 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-13 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-20 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-06 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-03 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-18 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-08 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-11 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-09 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-19 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-04 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-02 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-01 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-12 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-07 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-05 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-10 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-15 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 21 counted toward the lift figure. The other 1 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +27 percentage points is the difference between those two pass rates over the 21 comparable cases.
The per-case answers from this run were removed by the retention sweep, so the case table below shows the verdicts without the text either arm produced. The counts above were recorded at the time and are unaffected. Answers are now kept for 180 days.
Other measured skills in the registry, with their headline benchmark lift.