Imported from tiangong-ai/agent-skills (email-imap-full-fetch/SKILL.md). Install upstream with npx skills add tiangong-ai/agent-skills --skill email-imap-full-fetch. Copyright stays with the author.

Email IMAP Full Fetch

Core Goal

• Fetch one target email by stable message reference from IMAP.
• Enforce lookup order: HEADER Message-Id exact match first, then uid fallback.
• Download full raw MIME via BODY.PEEK[].
• Parse and return headers, full text body, html body, and attachment metadata.
• Save .eml and attachment files to disk with filename safety and idempotent indexing.

Standard Flow

1. Input must include message_id_norm from stage-1 routing output (mail_ref.message_id_norm).
2. Use fetch --message-id "<message_id_norm or raw Message-Id>" as the default path.
3. Use fetch --uid "<uid>" only when no usable message-id is available.
4. Keep mailbox selection consistent with stage-1 (--mailbox or IMAP_MAILBOX).
5. Read JSON output and continue downstream processing with returned mail_ref.

Commands

Fetch by Message-Id (preferred):

python3 scripts/imap_full_fetch.py fetch --message-id "<caa123@example.com>"

Fetch by UID (fallback only):

python3 scripts/imap_full_fetch.py fetch --uid "123456"

Use both when needed (message-id lookup first, uid fallback second):

python3 scripts/imap_full_fetch.py fetch --message-id "<caa123@example.com>" --uid "123456"

Output Contract

• Output is a single JSON object.
• Required top-level fields:
  • mail_ref
  • headers
  • text_plain
  • text_html
  • attachments
  • saved_eml_path
• mail_ref contains:
  • account, mailbox, uid, message_id_raw, message_id_norm, date
• attachments[] contains per-file metadata and persistence result:
  • filename, content_type, bytes, disposition, saved_path, skipped_reason

Storage And Idempotency

• saved_eml_path points to local .eml file saved from BODY.PEEK[].
• Attachments are saved without returning attachment binary content in JSON.
• Filenames are sanitized to remove path separators and unsafe characters.
• Duplicate attachment names are deduped with content-hash suffix.
• Repeated requests are idempotent by message_id_norm index and return existing persisted JSON record directly.

Parameters

• --message-id: primary lookup key.
• --uid: fallback lookup key.
• --mailbox: mailbox to query (default IMAP_MAILBOX or INBOX).
• --save-eml-dir: target dir for .eml files (env IMAP_FULL_SAVE_EML_DIR).
• --index-dir: target dir for idempotency index JSON files (env IMAP_FULL_INDEX_DIR, default <save-eml-dir>/.index).
• --save-attachments-dir: target dir for attachments (env IMAP_FULL_SAVE_ATTACHMENTS_DIR).
• --max-attachment-bytes: max saved attachment size (env IMAP_FULL_MAX_ATTACHMENT_BYTES).
• --allow-ext: allowed attachment extensions, comma-separated (env IMAP_FULL_ALLOW_EXT).
• --connect-timeout: IMAP connect timeout seconds (default from IMAP_CONNECT_TIMEOUT).

Required Environment

• IMAP_HOST
• IMAP_USERNAME
• IMAP_PASSWORD

Optional account defaults:

• IMAP_NAME
• IMAP_PORT
• IMAP_SSL
• IMAP_MAILBOX

Scripts

• scripts/imap_full_fetch.py