email-imap-fetch
Listen for one or more IMAP inboxes with the IDLE command, fetch unread email metadata plus text previews, and forward each message to OpenClaw webhooks. Use when tasks need near-real-time mailbox monitoring, multi-account inbox ingestion via environment variables, and automatic trigger delivery into OpenClaw automation.
Works with
--- name: email-imap-fetch description: Listen for one or more IMAP inboxes with the IDLE command, fetch unread email metadata plus text previews, and forward each message to OpenClaw webhooks. Use when tasks need near-real-time mailbox monitoring, multi-account inbox ingestion via environment variables, and automatic trigger delivery into OpenClaw automation. license: MIT --- # Email IMAP Fetch ## Core Goal - Wait for new mail with IMAP IDLE. - Fetch unread messages after each wake-up. - Support multiple mailbox accounts configured with env. - Control IDLE support strictly with env mode (`idle` or `poll`) without runtime probing. - Forward each fetched email to OpenClaw webhooks. - Emit machine-readable JSON lines for downstream steps. - Keep this skill strictly in stage-1 routing mode: send snippet + structured refs only, never send full raw message body, and never send attachment binary/content. ## Workflow 1. Configure account env variables and OpenClaw webhook env variables (see `references/env.md` and `assets/config.example.env`). 2. Validate configuration: ```bash python3 scripts/imap_idle_fetch.py check-config ``` 3. Run one IDLE cycle per account (smoke test): ```bash python3 scripts/imap_idle_fetch.py listen --cycles 1 --idle-seconds 120 --max-messages 10 ``` 4. Run continuously (default resident mode): ```bash python3 scripts/imap_idle_fetch.py listen ``` ## Runtime Model - Skill files are installed locally, but the listener is not auto-started. - In `idle` mode, IMAP IDLE receives push events only while listener process and IMAP connection are alive. - In `poll` mode, listener sleeps for poll interval and then fetches unread messages. - If the process exits, push events are missed; next run can still fetch existing unread emails with `UNSEEN`. - Default runtime is resident mode (`IMAP_CYCLES=0` by default). - Default IDLE mode is `poll` (safe for servers without IDLE support). - In production, always-on deployment must run under `systemd`, `launchd`, `supervisor`, or an equivalent daemon manager. - Do not run the listener as a foreground process bound to an interactive exec session; once that session exits, the listener will stop. ## Output Contract - Output format is JSONL (one JSON object per line). - `type=status` for lifecycle events. - `type=message` for fetched emails with: - `account`, `mailbox`, `seq`, `uid` - `subject`, `from`, `to`, `date` - `message_id_raw`, `message_id_norm` (and compatibility field `message_id`) - `snippet` (plain-text preview only) - `attachment_count`, `attachment_manifest` (summary only, no attachment content) - `mail_ref` machine-readable object (`account`, `mailbox`, `uid`, `message_id_raw`, `message_id_norm`, `date`) - Webhook message includes two fixed machine-readable blocks for deterministic dispatch extraction: - `<<<MAIL_REF_JSON>>> ... <<<END_MAIL_REF_JSON>>>` - `<<<ATTACHMENT_MANIFEST_JSON>>> ... <<<END_ATTACHMENT_MANIFEST_JSON>>>` - `wait_mode` is `idle` or `poll` in cycle status output. - `wait_events` records the active wait strategy details. - `event=webhook_delivered` status events when OpenClaw webhook POST succeeds. - `type=error` for account-level failures. - `event=webhook_failed` error events when OpenClaw webhook POST fails. ## Parameters - `--cycles`: IDLE cycles per account (`0` means forever). - `--idle-seconds`: max wait time for each IDLE call. - `--poll-seconds`: interval used when polling mode is active. - `--idle-mode`: `idle` or `poll`. - `--max-messages`: max unread emails fetched each cycle. - `--mark-seen` / `--no-mark-seen`: control unread state updates. - `--snippet-chars`: preview length limit. - `--connect-timeout`: connection timeout seconds. - `--retry-seconds`: retry delay after failure. Environment defaults: - `IMAP_CYCLES` - `IMAP_IDLE_MODE` - `IMAP_IDLE_SECONDS` - `IMAP_POLL_SECONDS` - `IMAP_MAX_MESSAGES` - `IMAP_MARK_SEEN` - `IMAP_SNIPPET_CHARS` - `IMAP_CONNECT_TIMEOUT` - `IMAP_RETRY_SECONDS` OpenClaw webhooks forwarding: - `OPENCLAW_WEBHOOKS_ENABLED` - `OPENCLAW_WEBHOOKS_TOKEN` - `OPENCLAW_WEBHOOKS_BASE_URL` - `OPENCLAW_WEBHOOKS_MODE` (`agent` or `wake`) - `OPENCLAW_WEBHOOKS_ENDPOINT` (optional endpoint override) - `OPENCLAW_WEBHOOKS_PATH` - `OPENCLAW_WEBHOOKS_WAKE_MODE` - `OPENCLAW_WEBHOOKS_DELIVER` - `OPENCLAW_WEBHOOKS_TIMEOUT` - `OPENCLAW_WEBHOOKS_NAME` - `OPENCLAW_WEBHOOKS_AGENT_ID` - `OPENCLAW_WEBHOOKS_CHANNEL` - `OPENCLAW_WEBHOOKS_TO` - `OPENCLAW_WEBHOOKS_MODEL` - `OPENCLAW_WEBHOOKS_THINKING` - `OPENCLAW_WEBHOOKS_AGENT_TIMEOUT_SECONDS` - `OPENCLAW_WEBHOOKS_SESSION_KEY_PREFIX` ## Error Handling - Invalid env configuration exits with code `2`. - In `idle` mode, unsupported IDLE returns explicit error and suggests `IMAP_IDLE_MODE=poll`. - Runtime failures are emitted as `type=error`. - Command exits non-zero when account processing errors occur. ## References - `references/env.md` ## Assets - `assets/config.example.env` ## Scripts - `scripts/imap_idle_fetch.py`
More API Design skills
lark-event
larksuite/cli
Lark/Feishu real-time event listening / subscribing / consuming: stream events as NDJSON via `lark-cli event consume <EventKey>` (covers IM messages/reactions/chat changes, Approval status changes, Task updates, VC meeting started/joined/ended, Minutes generated, Whiteboard updated, etc.). Use for Lark bots, real-time message processing, long-running subscribers, streaming webhook/push handlers. Supports `--max-events` / `--timeout` bounded runs and a stderr ready-marker contract — designed for AI agents running as subprocesses.
lark-contact
larksuite/cli
飞书 / Lark 通讯录:按姓名 / 邮箱解析成 open_id,或按 open_id 反查姓名 / 部门 / 邮箱 / 联系方式 / 个人状态 / 签名,以及按关键词搜索当前用户可见的机器人 / 智能体(agent)。当用户提到一个名字要下一步发消息 / 排日程,或拿到 open_id 想查具体信息时使用。不负责部门树遍历、按部门列员工、组织架构图,这类需求走原生 OpenAPI。
lark-openapi-explorer
larksuite/cli
飞书/Lark 原生 OpenAPI 探索:从官方文档库中挖掘未经 CLI 封装的原生 OpenAPI 接口。当用户的需求无法被现有 lark-* skill 或 lark-cli 已注册命令满足,需要查找并调用原生飞书 OpenAPI 时使用。

