[Silthus]

Problem

A Resend call site that sends a copy, a custom header, a tag or a file can't move to posthog.emails.send(): the code form takes only the content, the sender and replyTo.

This PR adds cc, bcc, headers, tags and attachments to the code form, as the shared SDK contract PostHog/posthog#120921 added them on 2026-10-22:

await posthog.emails.send({
  to: 'jane@example.com',
  distinctId: user.id,
  templateKey: 'invoice',
  subject: 'Your invoice',
  html: invoiceHtml,
  cc: 'finance@example.com',
  tags: [{ name: 'category', value: 'billing' }],
  attachments: [{ filename: 'invoice.pdf', content: await readFile('invoice.pdf') }],
})

Follow-up slice of strand Silthus/posthog#536, map Silthus/posthog#461.

Stacked on #5371 (code form), on #5312. #5372 (emails.get) now stacks on this PR. This fork PR targets main, so its diff includes the PRs below it. This layer alone: feat/node-emails-code-form...feat/node-emails-code-fields.

Server dependency, not merged: the fields come from PostHog/posthog#121851 (cc, bcc, headers, tags) and PostHog/posthog#121852 (attachments), stacked on #121272. This PR is built against the contract and those serializers at the HTTP seam.

Impact

Pays into Grow Workflows through billed workflows_emails from API sends, read as orgs that switched from Resend (not measured yet: the send attribution of PostHog/posthog#120628 plus its insight). In scope: Messaging, email. Expected move: small, up: it removes a blocker for Resend call sites that use these fields. Run cost: n/a.

Merge Danger

Caution

🔴 Merge risk: high · one-way door (merge-risk rubric)

• Adds cc, bcc, headers, tags and attachments to the public code form. Once released, the field names and attachment encoding are a contract, and a send delivers a real email, with its copies, to real recipients.
• The server checks every limit; the SDK checks types and refuses these fields on the template form. Attachment content is never logged. Tests at the HTTP boundary pin the exact body; not yet run against a live server (gap below).

Changes

Kind	Files	Lines (this layer)
Production	2	+120 −10
Tests	2	+131 −0
References and changeset	2	+102 −2

• Code form only. The new fields work with rendered content and with a local template. The template form's types refuse them, and at runtime a template send with one of them returns invalid_form on that field instead of sending without it.

• On the wire. Each field is left out when absent.

  Field	Request
  cc, bcc	A string becomes a one-item list
  headers, tags	As given
  attachments	{ filename, content, content_type?, content_id? }. A Buffer or Uint8Array becomes Base64; a string goes as is

• Once per send. The SDK encodes attachments when it builds the body, before the retry loop and before a local template's hash. Retries and the resend after a local-template upload send the same body. The upload never carries these fields.

• An older server. A server that doesn't take one of these fields yet answers 400 invalid_request on it; the error names it in field, so the caller sees which field to drop.

• A Resend path (a URL or file path to fetch) returns invalid_attachment, named validation_error, with field: attachments__<n>__path, before any request. So does an attachment without Buffer, Uint8Array or string content, or attachments that is not a list. The types refuse path too.

• Content stays private. The SDK never logs attachment content, and no error repeats it.

• Limits stay on the server. The SDK checks types only; the server refuses more than 10 copies, reserved headers, bad tag names, blocked file types and more than 512 KB.

Parity with the contract

Contract row	posthog-node (this PR)
cc, bcc: string or list	Always a list
headers: map	As is
tags: { name, value }[]	As is
attachments: { filename, content, content_type?, content_id? }	content Base64 from a Buffer or Uint8Array, a string as is
Template form	Refused by the types and at runtime with invalid_form
Attachment content	Never logged or repeated, encoded once before the retry loop

Evidence

• Test seams:
  • The emails namespace on a real client, fetch mocked by FakeEmailsApi (emails.spec.ts).
  • The form types, checked by tsc with @ts-expect-error cases (__tests__/types/emails-forms.ts).
• What the tests pin:
  • The exact body of a rendered send with every new field. It covers a Buffer, a Uint8Array, a Base64 string, and a 40 KB file that crosses the encoder's chunk size.
  • A local template that sends the fields on both sends, with an identical body, and never in its upload.
  • invalid_attachment without a request, for a path, for content of another type, and for an attachments object that is not a list.
  • invalid_form with its field for each new field on the template form, without a request.
  • 400 invalid_request named validation_error, with its field.
  • Types:
    • The template form refuses cc and attachments.
    • An attachment refuses path.
• Red first: 4 cases and the type tests failed before the change, then green. The review fix's 5 cases failed before it, then green.
• Local gate on abeb1979 (on the leased compute box): node vitest (1336 passed), the edge build tests (1336 passed), tsc for the build and the type tests, oxlint, oxfmt, build and generate-references; core vitest.
• Owed proofs:
  • Live run (a send with a copy and an attachment into the dev mail catcher, and an older server's invalid_request): gap. feat(messaging): send cc, bcc, headers and tags with a code send posthog#121851 and feat(messaging): send attachments with a code send posthog#121852 are drafts being restacked; the end-to-end lane runs it.

Release info Sub-libraries affected

Libraries affected

• All of them
• posthog-js (web)
• posthog-js-lite (web lite)
• posthog-node
• posthog-react-native
• @posthog/react-native-plugin
• @posthog/react
• @posthog/ai
• @posthog/convex
• @posthog/next
• @posthog/nextjs-config
• @posthog/nuxt
• @posthog/openfeature-node-provider
• @posthog/openfeature-web-provider
• @posthog/rollup-plugin
• @posthog/webpack-plugin
• @posthog/types
• @posthog/browser-common

Checklist

• Tests for new code
• Accounted for the impact of any changes across different platforms
• Accounted for backwards compatibility of any changes (no breaking changes!)
• Took care not to unnecessarily increase the bundle size

If releasing new changes

• Ran pnpm changeset to generate a changeset file

🤖 Agent context

Autonomy: Human-driven (agent-assisted)

Michael (@Silthus) approved adding the fields to the contract and to posthog-node. Built by Claude Code (Claude Opus 5.5) test-first, as one build lane of the map.

Question	Answer	Who	Where
Which fields the code form adds	cc, bcc, headers, tags, attachments, code form only	👤 Human	#120921, Changes
Attachment content types	Buffer, Uint8Array or a Base64 string, snake_case on the wire	👤 Human	#120921, Client interface
Resend's attachment path	Refused with invalid_attachment before any request	👤 Human	lane brief for this slice
Where this PR sits in the stack	Between #5371 and #5372, so the stack stays one line	🤖 Agent	this PR
Base64 encoder	btoa over byte chunks, not Buffer, so the edge build works too	🤖 Agent	this PR

Open: none.

Involvement: decisions by Michael through the contract; one review by GPT-6.1 Sol with no findings; a QA swarm over the restacked stack in two rounds (Review lane GPT-6.1 Sol with seven lenses, Fresh-eyes lane Claude Opus 5.5) with its fix in this PR; no live run yet.

🤖 Generated with Claude Code