agentvibes 5.6.8 → 5.7.0

package/.clawdbot/README.md

@@ -1,105 +1,105 @@
-# Clawdbot Integration
-
-AgentVibes TTS integration for Clawdbot - the AI assistant framework.
-
-## What is Clawdbot?
-
-Clawdbot is a powerful AI assistant framework that connects Claude AI to messaging platforms (WhatsApp, Telegram, Discord, etc.) and provides extensible skills.
-
-**Website**: https://clawd.bot
-**GitHub**: https://github.com/clawdbot/clawdbot
-
-## AgentVibes + Clawdbot
-
-This integration provides Clawdbot with professional TTS capabilities using AgentVibes:
-
-- **Free & Offline**: No API costs, works without internet
-- **50+ Voices**: Professional AI voices in 30+ languages
-- **Remote SSH Support**: Audio tunnels from server to local machine
-- **Zero Setup**: Automatic when AgentVibes is installed
-
-## Installation
-
-```bash
-npx agentvibes install
-```
-
-That's it! AgentVibes is ready to use.
-
-## Usage from Clawdbot
-
-Once installed, use AgentVibes for TTS:
-
-```bash
-npx agentvibes speak "Hello from Clawdbot"
-```
-
-## Features
-
-- **Automatic Detection**: Clawdbot automatically uses AgentVibes when available
-- **Voice Selection**: 50+ voices across 30+ languages
-- **SSH Tunneling**: Audio plays on local machine from remote servers
-- **Zero Config**: Works out of the box once AgentVibes is installed
-
-## Remote SSH Audio
-
-When running Clawdbot on a remote server, set up PulseAudio tunneling so audio plays on your local machine:
-
-**Quick Setup:**
-
-1. **Remote server** (`~/.bashrc`):
-```bash
-export PULSE_SERVER=tcp:localhost:14713
-```
-
-2. **Local machine** (`~/.ssh/config`):
-```
-Host your-server
-    RemoteForward 14713 localhost:14713
-```
-
-3. **Connect and test**:
-```bash
-ssh your-server
-agentvibes speak "Testing remote audio"
-```
-
-**Full guide**: https://github.com/paulpreibisch/AgentVibes/blob/master/docs/remote-audio-setup.md
-
-## Documentation
-
-- **Skill Documentation**: [skill/SKILL.md](skill/SKILL.md)
-- **AgentVibes Docs**: https://agentvibes.org
-- **Clawdbot Docs**: https://docs.clawd.bot
-
-## Architecture
-
-```
-Clawdbot Request
-    ↓
-AgentVibes Skill (this)
-    ↓
-AgentVibes CLI (agentvibes speak)
-    ↓
-Piper TTS Engine
-    ↓
-PulseAudio → SSH Tunnel → Local Speakers
-```
-
-## Development
-
-The skill is automatically packaged with AgentVibes npm releases. To update:
-
-1. Edit `skill/SKILL.md`
-2. Test with Clawdbot
-3. Commit to AgentVibes repo
-4. Publish new AgentVibes version
-
-## Support
-
-- **AgentVibes Issues**: https://github.com/paulpreibisch/AgentVibes/issues
-- **Clawdbot Issues**: https://github.com/clawdbot/clawdbot/issues
-
-## License
-
-Apache 2.0 - Same as AgentVibes
+# Clawdbot Integration
+
+AgentVibes TTS integration for Clawdbot - the AI assistant framework.
+
+## What is Clawdbot?
+
+Clawdbot is a powerful AI assistant framework that connects Claude AI to messaging platforms (WhatsApp, Telegram, Discord, etc.) and provides extensible skills.
+
+**Website**: https://clawd.bot
+**GitHub**: https://github.com/clawdbot/clawdbot
+
+## AgentVibes + Clawdbot
+
+This integration provides Clawdbot with professional TTS capabilities using AgentVibes:
+
+- **Free & Offline**: No API costs, works without internet
+- **50+ Voices**: Professional AI voices in 30+ languages
+- **Remote SSH Support**: Audio tunnels from server to local machine
+- **Zero Setup**: Automatic when AgentVibes is installed
+
+## Installation
+
+```bash
+npx agentvibes install
+```
+
+That's it! AgentVibes is ready to use.
+
+## Usage from Clawdbot
+
+Once installed, use AgentVibes for TTS:
+
+```bash
+npx agentvibes speak "Hello from Clawdbot"
+```
+
+## Features
+
+- **Automatic Detection**: Clawdbot automatically uses AgentVibes when available
+- **Voice Selection**: 50+ voices across 30+ languages
+- **SSH Tunneling**: Audio plays on local machine from remote servers
+- **Zero Config**: Works out of the box once AgentVibes is installed
+
+## Remote SSH Audio
+
+When running Clawdbot on a remote server, set up PulseAudio tunneling so audio plays on your local machine:
+
+**Quick Setup:**
+
+1. **Remote server** (`~/.bashrc`):
+```bash
+export PULSE_SERVER=tcp:localhost:14713
+```
+
+2. **Local machine** (`~/.ssh/config`):
+```
+Host your-server
+    RemoteForward 14713 localhost:14713
+```
+
+3. **Connect and test**:
+```bash
+ssh your-server
+agentvibes speak "Testing remote audio"
+```
+
+**Full guide**: https://github.com/paulpreibisch/AgentVibes/blob/master/docs/remote-audio-setup.md
+
+## Documentation
+
+- **Skill Documentation**: [skill/SKILL.md](skill/SKILL.md)
+- **AgentVibes Docs**: https://agentvibes.org
+- **Clawdbot Docs**: https://docs.clawd.bot
+
+## Architecture
+
+```
+Clawdbot Request
+    ↓
+AgentVibes Skill (this)
+    ↓
+AgentVibes CLI (agentvibes speak)
+    ↓
+Piper TTS Engine
+    ↓
+PulseAudio → SSH Tunnel → Local Speakers
+```
+
+## Development
+
+The skill is automatically packaged with AgentVibes npm releases. To update:
+
+1. Edit `skill/SKILL.md`
+2. Test with Clawdbot
+3. Commit to AgentVibes repo
+4. Publish new AgentVibes version
+
+## Support
+
+- **AgentVibes Issues**: https://github.com/paulpreibisch/AgentVibes/issues
+- **Clawdbot Issues**: https://github.com/clawdbot/clawdbot/issues
+
+## License
+
+Apache 2.0 - Same as AgentVibes

package/.mcp.json

@@ -1,21 +1,30 @@
 {
   "mcpServers": {
-    "firecrawl-mcp": {
+    "agentvibes": {
       "command": "npx",
       "args": [
         "-y",
+        "--package=agentvibes",
+        "agentvibes-mcp-server"
+      ],
+      "env": {
+        "AGENTVIBES_MCP_FALLBACK": "copilot"
+      }
+    },
+    "firecrawl": {
+      "command": "npx",
+      "args": [
         "firecrawl-mcp"
       ]
     },
-    "agentvibes": {
+    "context7": {
       "command": "npx",
       "args": [
         "-y",
-        "--package=agentvibes",
-        "agentvibes-mcp-server"
+        "@upstash/context7-mcp"
       ],
       "env": {
-        "AGENTVIBES_MCP_FALLBACK": "copilot"
+        "CONTEXT7_API_KEY": "${CONTEXT7_API_KEY}"
       }
     }
   }

package/README.md

@@ -11,7 +11,7 @@
 [![Publish](https://github.com/paulpreibisch/AgentVibes/actions/workflows/publish.yml/badge.svg)](https://github.com/paulpreibisch/AgentVibes/actions/workflows/publish.yml)
 [![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
 
-**Author**: Paul Preibisch ([@997Fire](https://x.com/997Fire)) | **Version**: v5.6.8
+**Author**: Paul Preibisch ([@997Fire](https://x.com/997Fire)) | **Version**: v5.7.0
 
 ---
 
@@ -40,7 +40,15 @@ Whether you're coding in Claude Code, chatting in Claude Desktop, using Warp Ter
 
 ---
 
-## 🌟 NEW IN v5.6.8 — WSL Voice Routing Fixed + Session Lifecycle Reliability
+## 🌟 NEW IN v5.7.0 — BMAD v6.6 Support + Windows Auto-Restart Watcher
+
+**BMAD v6.6.0:** AgentVibes now detects the new `.claude/skills/*/agents/` agent structure, correctly handles globally-installed BMAD at `~/_bmad`, and gracefully skips v6.6+ plain-Markdown agents during TTS injection instead of erroring. The BMAD tab now shows detection correctly for global installs.
+
+**Windows watcher:** `tts-watcher.ps1` is now a standalone file at `~/.agentvibes/tts-watcher.ps1`. Running `npx agentvibes update` now copies the latest watcher **and** restarts it automatically — both the file and the process are updated in one step, no manual restart needed.
+
+**Windows provider:** `play-tts.ps1` now respects the `ProviderOverride` from the Linux server config when receiving remote audio.
+
+## v5.6.9 — Reverb & Background Music Silent in NPX Installs
 
 **WSL users:** AgentVibes was playing `en_US-lessac-medium` regardless of your configured voice. Fixed — Piper is now found in non-interactive shells by explicitly prepending `~/.local/bin` to `PATH` before the binary check.
 

package/RELEASE_NOTES.md

@@ -1,5 +1,66 @@
 # AgentVibes Release Notes
 
+## 🎭 v5.7.0 — BMAD v6.6 Support + Windows Auto-Restart Watcher
+
+**Released:** 2026-05-11
+
+### 🆕 BMAD v6.6.0 Compatibility
+
+BMAD v6.6 restructured where agents live — they moved from `_bmad/bmm/agents/` to `.claude/skills/*/agents/`. AgentVibes now detects and scans these new paths correctly.
+
+**TTS injection** gracefully skips v6.6+ agents (which use plain Markdown without XML/YAML activation sections) instead of throwing errors. The install summary now clearly reports how many agents were skipped vs. modified.
+
+**BMAD tab detection** now finds globally-installed BMAD at `~/_bmad` (home-dir install) in addition to project-local installs. Previously the BMAD tab showed "Not detected" even when BMAD was installed globally.
+
+**Security:** The installer's path validation now correctly permits BMAD paths under the user's home directory, fixing a false-positive "Invalid BMAD path" error for global installs.
+
+### 🆕 Windows TTS Watcher — Standalone File + Auto-Restart
+
+`tts-watcher.ps1` is now extracted to `~/.agentvibes/tts-watcher.ps1` as a standalone file. Running `npx agentvibes update` now copies the latest watcher to that location **and** automatically restarts it — so both the watcher script itself and its running process are updated in one step. No manual file replacement or restart needed after updates.
+
+### 🐛 Windows Provider Override Respected on Laptop
+
+`play-tts.ps1` now reads the `ProviderOverride` setting from the Linux-side config when receiving audio via SSH. Previously the laptop always used its locally-configured provider even if the server specified a different one.
+
+### 🐛 Sample Command Added to Voice Manager
+
+`voice-manager.sh sample` was missing its handler — calling it fell through to the usage/exit path silently. Fixed.
+
+### 🐛 Preview SSH Routing Detects Correct Endpoint
+
+`provider-manager.sh` now includes `detect_routing_llm()` which checks `AGENTVIBES_LLM_KEY` then `transport-config.json` for the first `mode=remote` entry, so preview audio reaches the correct SSH host.
+
+---
+
+## 🔇 v5.6.9 — Reverb & Background Music Silent in NPX Installs
+
+**Released:** 2026-05-09
+
+### 🐛 Reverb and Background Music Silently Broken for All NPX Users
+
+When AgentVibes is installed via `npx`, hook files are extracted from the tarball with 644 permissions — no execute bit. `play-tts-piper.sh` called `audio-processor.sh` directly, which exits immediately with code 126 (Permission denied) on a non-executable file. Every `npx`-installed user was getting voice-only TTS — no reverb, no background music, silently.
+
+**Fix 1:** `play-tts-piper.sh` now calls `audio-processor.sh` via `bash "$SCRIPT_DIR/audio-processor.sh"`, bypassing the execute-bit check.
+**Fix 2:** `install-deps.js` (postinstall) now runs `ensureHookPermissions()` to `chmod 755` all `.sh` files after npm install.
+
+### 🐛 Voice Browser Preview Ignored Reverb and Background Music
+
+The **Preview** button in the Voice Browser played raw piper output with no reverb and no background music, bypassing `audio-processor.sh` entirely.
+
+**Fix:** Preview audio now routes through the same `audio-processor.sh` pipeline as real TTS.
+
+### 🐛 MCP `text_to_speech` Returned Garbled File Path and Missing Voice Info
+
+The tool extracted the audio file path incorrectly (trailing size/emoji characters included) and never reported the voice name in its response.
+
+**Fix:** ANSI codes are stripped before parsing, the `.wav` path is cleanly extracted, and the `🎤 Voice used:` line is included in the tool response.
+
+### 🐛 Background Music TUI Toggle Didn't Take Effect
+
+Enabling background music in the **Music** tab wrote to `config.json` but not to `background-music-enabled.txt` (read by bash hooks). Music stayed off after toggling. Saving a track also now implies enabling music.
+
+---
+
 ## 🐧 v5.6.8 — WSL Voice Routing Fixed + Session Lifecycle Reliability
 
 **Released:** 2026-05-09