Response Schemas

Response schemas are easy to ignore — but they pay back disproportionately. Two reasons: they make serialization roughly 3x faster and they keep sensitive fields out of the wire payload.

Status-Keyed Responses

const userSchema = {
  type: 'object',
  properties: {
    id: { type: 'integer' },
    email: { type: 'string', format: 'email' },
    createdAt: { type: 'string', format: 'date-time' },
  },
  required: ['id', 'email'],
} as const;

app.get('/me', {
  schema: {
    response: {
      200: userSchema,
      401: {
        type: 'object',
        properties: { message: { type: 'string' } },
      },
    },
  },
  handler: async (req) => {
    return db.users.findCurrent(req); // returns { id, email, passwordHash, ... }
  },
});

Even though the row carries passwordHash, the serializer only writes id, email, and createdAt to the response. No more manually picking columns or remembering to scrub fields.

Why It’s Faster

Fastify uses fast-json-stringify, which compiles the schema into a hand-rolled JS function at boot. There’s no per-call schema walk and no type checks — it knows the layout in advance. Hot endpoints see 2-3x throughput improvements.

Match Multiple Status Codes

Use the 2xx, 3xx, etc. wildcards when the shape repeats:

response: {
  '2xx': userSchema,
  '4xx': { type: 'object', properties: { message: { type: 'string' } } },
}

If you return a status not covered by any key, Fastify falls back to JSON.stringify.