[stevenobiajulu]

Closes #214.

What

create_draft and update_draft now accept bcc: string[] (bare or name-address form), on new drafts and on the reply path. Microsoft Graph writes bccRecipients; Gmail writes a Bcc header. bcc is also a recognized body-file frontmatter key. Providers, ComposeMessage/ReplyOptions, the draft preview, and the send_draft allowlist gate already carried bcc, so most of this is the input surface.

Behavior choices worth reviewing

• Reply path: Bcc wins over derived To/Cc. The motivating case is "moving you to bcc" on a reply-all. Without this, Graph's createReplyAll (and Gmail's derived recipients) would keep the introducer on To/Cc and add them to Bcc, so they'd stay visible. Both providers now remove Bcc addresses (case-insensitive) from the To/Cc they derive. Gmail omits the To: header if that empties it. Graph only patches toRecipients/ccRecipients when something actually has to come off, so the no-bcc path sends the same PATCH as before.
• Explicit conflict is rejected. The same address on bcc and to/cc in one call returns BCC_RECIPIENT_CONFLICT (recoverable).
• update_draft read-back warning. update_draft with only bcc can't see the draft's existing To/Cc. If the persisted preview still shows a Bcc address on To/Cc, the response carries a BCC_RECIPIENT_ALSO_VISIBLE warning. bcc replaces the Bcc line, [] clears it, and omitting it keeps it.
• send_email refuses bcc frontmatter (BCC_NOT_SUPPORTED). It shares resolveComposeFields, has no bcc input, and its allowlist covers only to/cc. Without the refusal, a body file meant for create_draft would send with no Bcc and no error.
• Allowlist unchanged. Drafts stay allowlist-exempt. send_draft already gates stored To/Cc/Bcc. New tests cover bcc set through both create_draft and update_draft being blocked at send time.

Tests

• Core (draft.test.ts, new Draft Bcc Recipients (issue #214) block): new draft, omitted bcc, reply-all move-to-bcc, derived-recipient removal, conflict, invalid address, frontmatter, the send_email refusal, update set/keep/clear, the visibility warning, and the send_draft allowlist for both create and update paths.
• frontmatter.test.ts: bcc list and single-value parsing.
• Gmail: Bcc header emitted and removed from derived To/Cc; empty To omitted.
• Graph: derived To/Cc filtering, Cc cleared to [] when needed, explicit to + bcc keeps the displaced participant off Cc.

npm run build, npm run lint, the full workspace test:run (all green), check:spec-coverage (290/290), and openspec validate add-draft-bcc --strict all pass.

OpenSpec change: openspec/changes/add-draft-bcc.

Not done

No live real-mailbox smoke yet. This touches both providers' reply-draft recipient handling, so it's worth creating a reply-all draft with bcc against a real Graph and Gmail mailbox (draft only, no send) and checking it in the client before release.

[codecov]

Codecov Report

❌ Patch coverage is 98.64865% with 1 line in your changes missing coverage. Please review.
✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
...ackages/provider-gmail/src/email-gmail-provider.ts	87.50%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!