Update requirements and enhance documentation for service installation and message handling

- Changed `meshcore` dependency version in `requirements.txt` from `>=2.2.14` to `==2.2.14` for consistency.
- Added detailed instructions in `SERVICE-INSTALLATION.md` regarding Python version compatibility and potential f-string issues with Python 3.11.
- Introduced a new section in `local-plugins.md` explaining how to handle message chunking and bot rate limits, providing code examples for better clarity.

docs/SERVICE-INSTALLATION.md

@@ -5,7 +5,7 @@ This guide explains how to install the MeshCore Bot as a systemd service on Linu
 ## Prerequisites
 
 - Linux system with systemd
-- Python 3.7+
+- Python 3.7+ (Python 3.12+ recommended; on 3.11 the meshcore dependency has an f-string bug — the install script patches it automatically)
 - Root/sudo access
 - MeshCore-compatible device
 
@@ -144,6 +144,16 @@ sudo systemctl restart meshcore-bot
 3. Check configuration: `sudo nano /opt/meshcore-bot/config.ini`
 4. Verify dependencies: `sudo pip3 list | grep meshcore`
 
+### SyntaxError: f-string: unmatched '[' (Python 3.11)
+If the bot fails on import with this error in `meshcore/commands/contact.py`, you are on Python 3.11 and the **meshcore** dependency uses an f-string that only works on Python 3.12+.
+
+**Options:**
+- **Recommended:** Use Python 3.12+ (create the venv with `python3.12` if available, then re-run `./install-service.sh --upgrade`).
+- **Or:** Re-run the install script so it can patch the installed package:  
+  `sudo ./install-service.sh --upgrade`  
+  The script detects Python 3.11 and patches the meshcore file in the venv.
+- **Manual patch:** Edit `/opt/meshcore-bot/venv/lib/python3.11/site-packages/meshcore/commands/contact.py`, find the line containing `contact["adv_name"]` inside the f-string, and change it to `contact['adv_name']` (single quotes around `adv_name`).
+
 ### Permission Issues
 1. Check file ownership: `ls -la /opt/meshcore-bot/`
 2. Fix permissions: `sudo chown -R meshcore:meshcore /opt/meshcore-bot`

docs/local-plugins.md

@@ -80,6 +80,37 @@ Restart the bot so the service is loaded and started.
 
 If a local command or service has the same **name** as an already-loaded plugin or service (e.g. you add `local/commands/ping.py` with `name = "ping"`), the local one is **skipped** and a warning is logged. Choose a different name (e.g. `pinglocal`) to avoid the conflict.
 
+## Sending multiple messages (chunking)
+
+When a service plugin sends a long message by splitting it into chunks and calling `send_channel_message` multiple times, the bot’s **rate limiters** can block the second and later sends. You’ll see a warning like “Rate limited. Wait X seconds.”
+
+**What’s going on**
+
+- **Global rate limit** (`[Bot]` `rate_limit_seconds`, default 10): minimum time between *any* two bot replies. If you don’t skip it, the first send uses the “slot” and the next send within that window is blocked.
+- **Bot TX rate limit** (`bot_tx_rate_limit_seconds`, default 1.0): minimum time between bot transmissions on the mesh. This is always enforced.
+
+**What to do**
+
+1. **Use `skip_user_rate_limit=True`** for every chunk. That skips the global (and per-user) limits so automated service messages aren’t blocked by the “10 second” global window.
+2. **Space chunks in time** so the bot TX limit is satisfied: before each chunk after the first, wait for the bot TX rate limiter and then sleep. Same pattern as the greeter and other multi-part senders:
+
+```python
+import asyncio
+
+# chunks = ["first part...", "second part...", ...]
+for i, chunk in enumerate(chunks):
+    if i > 0:
+        await self.bot.bot_tx_rate_limiter.wait_for_tx()
+        rate_limit = self.bot.config.getfloat('Bot', 'bot_tx_rate_limit_seconds', fallback=1.0)
+        sleep_time = max(rate_limit + 0.5, 1.0)
+        await asyncio.sleep(sleep_time)
+    await self.bot.command_manager.send_channel_message(
+        self.channel, chunk, skip_user_rate_limit=True
+    )
+```
+
+So you are allowed to send multiple messages in sequence; you do **not** need 10 seconds between chunks. Use `skip_user_rate_limit=True` and about 1–1.5 seconds (or your configured `bot_tx_rate_limit_seconds` + buffer) between chunks.
+
 ## References
 
 - [Service plugins](service-plugins.md) — built-in services and how they are enabled.

requirements.txt

@@ -18,7 +18,7 @@ pytz>=2023.3
 aiohttp>=3.8.0
 cryptography>=41.0.0
 pynacl>=1.5.0
-meshcore>=2.2.14
+meshcore==2.2.14
 openmeteo-requests>=1.7.2
 requests-cache>=1.1.1
 retry-requests>=1.0.0