feat(gateway): feishu voice message STT via gateway audio attachment

docs/feishu.md

@@ -80,6 +80,8 @@ https://your-gateway-host/webhook/feishu
 | — | `FEISHU_ALLOW_BOTS` | `off` | Bot message handling: `off` / `mentions` / `all` |
 | — | `FEISHU_TRUSTED_BOT_IDS` | — | Comma-separated open_id list of known bots |
 | — | `FEISHU_MAX_BOT_TURNS` | `20` | Max consecutive bot replies per channel before suppression |
+| — | `FEISHU_SESSION_TTL_HOURS` | `24` | How long the bot remembers thread participation (hours). After expiry, @mention is required again. |
+| — | `FEISHU_ALLOW_USER_MESSAGES` | `involved` | Thread response mode: `involved` / `mentions` / `multibot-mentions`. See below. |
 | `gateway.botUsername` | — | — | Set to bot's `open_id` for @mention gating |
 | `gateway.streaming` | — | `false` | Enable streaming (typewriter) mode |
 
@@ -95,6 +97,32 @@ In group chats, the bot only responds when @mentioned (default). To find your bo
 
 To disable mention gating: `feishu.requireMention: false`.
 
+### Thread Participation (Involved Mode)
+
+Once the bot replies in a thread (topic), it remembers that thread and responds to subsequent messages **without requiring @mention** — similar to Discord's `allow_user_messages: "involved"` mode.
+
+- Only applies to threads (messages with `root_id`). Main channel messages always require @mention.
+- Participation is stored in memory. Gateway restart clears the cache; users need to @mention once to re-engage.
+- TTL controlled by `FEISHU_SESSION_TTL_HOURS` (default 24h). After expiry, @mention is required again.
+
+### Multi-Bot Threads (multibot-mentions Mode)
+
+When `FEISHU_ALLOW_USER_MESSAGES=multibot-mentions`, the bot detects when another bot is @mentioned in a participated thread and reverts to requiring @mention — preventing all bots from responding simultaneously.
+
+| Mode | Behavior |
+|------|----------|
+| `involved` (default) | Bot responds in participated threads without @mention. All participated bots respond. |
+| `multibot-mentions` | Same as `involved`, but once another bot is @mentioned in the thread, require @mention for all bots. |
+| `mentions` | Always require @mention, even in participated threads. |
+
+**Multi-bot detection** (how the gateway identifies "another bot"):
+
+1. If `FEISHU_TRUSTED_BOT_IDS` is set → exact match against configured IDs
+2. If only `FEISHU_ALLOWED_USERS` is set → any @mention that is not self and not in allowed_users is inferred as another bot (recommended, zero-config)
+3. If neither is set → no multibot detection
+
+Note: Detection only triggers in threads where the bot has already participated. This prevents premature marking of threads the bot hasn't joined.
+
 ## Security Notes
 
 - `appSecret`, `verificationToken`, and `encryptKey` are stored in a Kubernetes Secret, not in ConfigMap.
@@ -139,6 +167,7 @@ The gateway downloads and forwards image and text file attachments to the AI age
 | `text` | Text extracted, forwarded as prompt |
 | `image` | Image downloaded, resized (max 1200px), JPEG compressed, base64 encoded → `ContentBlock::Image` |
 | `file` | Text files only (`.txt`, `.py`, `.rs`, `.md`, `.json`, etc., max 512KB). Non-text files (`.pdf`, `.zip`, etc.) are silently ignored. |
+| `audio` | Voice message downloaded (opus/ogg, max 25MB), base64 encoded, forwarded to core. If `[stt]` is enabled, core transcribes via Whisper API and injects `[Voice message transcript]: ...` into the prompt. If STT is disabled or fails, the message is silently skipped. |
 | `post` | Rich text: text nodes extracted as prompt, `img` nodes downloaded as image attachments. This is the format Feishu uses when @mention + paste image in a group. |
 
 **Group chat limitation:** Feishu does not allow @mention and image upload in the same message. However, @mention + paste (Ctrl+V) an image works — Feishu sends this as a `post` message containing both the mention and the image. Direct image upload (via the attachment button) cannot include @mention, so the bot will not respond in groups.

docs/stt.md

@@ -1,6 +1,6 @@
 # Speech-to-Text (STT) for Voice Messages
 
-openab can automatically transcribe Discord voice message attachments and forward the transcript to your ACP agent as text.
+openab can automatically transcribe voice message attachments (Discord, Feishu, and other gateway platforms) and forward the transcript to your ACP agent as text.
 
 ## Quick Start
 
@@ -24,7 +24,7 @@ api_key = "${GROQ_API_KEY}"
 ## How It Works
 
 ```
-Discord voice message (.ogg)
+Voice message (Discord .ogg, Feishu opus/ogg, etc.)
        │
        ▼
   openab downloads the audio file
@@ -161,6 +161,6 @@ When disabled, audio attachments are silently skipped with no impact on existing
 ## Technical Notes
 
 - openab sends `response_format=json` in the transcription request to ensure the response is always parseable JSON. Some local whisper servers default to plain text output without this parameter.
-- The actual MIME type from the Discord attachment is passed through to the STT API (e.g. `audio/ogg`, `audio/mp4`, `audio/wav`).
+- The actual MIME type from the platform attachment is passed through to the STT API (e.g. `audio/ogg` for Discord and Feishu voice messages, `audio/mp4`, `audio/wav`).
 - Environment variables in config values are expanded via `${VAR}` syntax (e.g. `api_key = "${GROQ_API_KEY}"`).
 - The `api_key` field is auto-detected from the `GROQ_API_KEY` environment variable when using the default Groq endpoint. If you set a custom `base_url` (e.g. local server), auto-detect is disabled to avoid leaking the Groq key to unrelated endpoints — you must set `api_key` explicitly.