feat: add QMD integration (Dockerfile, agent-runner, container skill)

Install QMD CLI in container, wire MCP server into agent-runner,
and add container skill for conversation search.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

container/Dockerfile

@@ -30,8 +30,8 @@ RUN apt-get update && apt-get install -y \
 ENV AGENT_BROWSER_EXECUTABLE_PATH=/usr/bin/chromium
 ENV PLAYWRIGHT_CHROMIUM_EXECUTABLE_PATH=/usr/bin/chromium
 
-# Install agent-browser and claude-code globally
+# Install agent-browser, claude-code, and QMD globally
-RUN npm install -g agent-browser @anthropic-ai/claude-code
+RUN npm install -g agent-browser @anthropic-ai/claude-code @tobilu/qmd
 
 # Create app directory
 WORKDIR /app

container/agent-runner/src/index.ts

@@ -469,6 +469,7 @@ async function runQuery(
         'Skill',
         'NotebookEdit',
         'mcp__nanoclaw__*',
+        'mcp__qmd__*',
       ],
       env: sdkEnv,
       permissionMode: 'bypassPermissions',
@@ -484,6 +485,10 @@ async function runQuery(
             NANOCLAW_IS_MAIN: containerInput.isMain ? '1' : '0',
           },
         },
+        qmd: {
+          type: 'http',
+          url: 'http://host.docker.internal:8182/mcp',
+        },
       },
       hooks: {
         PreCompact: [

container/skills/qmd/SKILL.md

@@ -0,0 +1,71 @@
+---
+name: qmd
+description: Search past conversations and documentation. Use when users ask about things mentioned before, past discussions, or need context from history.
+allowed-tools: Bash(npx qmd:*), Grep, Glob, Read
+---
+
+# QMD - Conversation Search
+
+Search past conversations and documentation in the groups directory.
+
+## MCP Tools (Preferred)
+
+QMD MCP server runs on the host at `http://host.docker.internal:8182/mcp`.
+
+Available tools:
+- `mcp__qmd__query` - Search with lex/vec/hyde queries
+- `mcp__qmd__get` - Retrieve document by path or docid
+- `mcp__qmd__multi_get` - Batch retrieve by glob pattern
+- `mcp__qmd__status` - Check index health
+
+Example query:
+```json
+{
+  "searches": [
+    { "type": "lex", "query": "search term" },
+    { "type": "vec", "query": "natural language question" }
+  ],
+  "collections": ["telegram_main"],
+  "limit": 10
+}
+```
+
+## CLI Fallback
+
+If MCP tools are unavailable, use the QMD CLI directly:
+
+```bash
+# Keyword search
+npx qmd search "search term" -c telegram_main
+
+# Semantic search (requires embeddings)
+npx qmd vsearch "natural language question" -c telegram_main
+
+# Hybrid search with reranking (best quality)
+npx qmd query "question" -c telegram_main
+```
+
+## Fallback: Direct File Search
+
+If QMD isn't available at all, search conversation files directly:
+
+```bash
+# Find conversations containing a term
+grep -r "term" /workspace/group/conversations/
+
+# List recent conversations
+ls -lt /workspace/group/conversations/ | head -10
+```
+
+## Conversation Files Location
+
+- Conversations: `/workspace/group/conversations/*.md`
+- Documentation: `/workspace/group/docs/*.md`
+- Group memory: `/workspace/group/CLAUDE.md`
+
+## When to Use
+
+- User asks "what did we discuss about X"
+- User mentions something from a past conversation
+- Need context from previous sessions
+- Looking up decisions or preferences mentioned before