fix: browser integration for AI research agent

- Dockerfile: Install Google Chrome stable instead of Playwright npm package
  (playwright-cli defaults to /opt/google/chrome/chrome path). Removed unused
  Playwright npm package, kept Playwright Python for page automation.

- bundle/bundle.md: Fix orchestrator config override with correct key path
  (session.orchestrator.config.extended_thinking: false) to properly override
  foundation's extended_thinking: true default, which was causing empty
  AI text responses.

- bundle/agents/researcher.md: Add strong pre-installed environment guardrail
  telling agent NOT to install packages. Add explicit bash() tool call examples
  showing correct playwright-cli syntax. Fix snapshot command (removed non-existent
  -ic flag), change 'type' to 'fill' for form field syntax.

- bundle/context/researcher-instructions.md: Enrich from thin pointer to
  concrete examples with correct bash() wrapping and actual playwright-cli
  commands. Add 'do NOT install' guardrail.

Validated: Agent now opens Chrome browser during research queries, noVNC panel
shows live browser activity, AI generates text responses properly.

Generated with Amplifier

Co-Authored-By: Amplifier <240397093+microsoft-amplifier@users.noreply.github.com>

Dockerfile

@@ -33,13 +33,14 @@ RUN curl -fsSL https://deb.nodesource.com/setup_22.x | bash - \
     && apt-get install -y --no-install-recommends nodejs \
     && rm -rf /var/lib/apt/lists/*
 
-# ── npm global tools: @playwright/cli + Chromium browser ──────────────────────
+# ── npm global tools: @playwright/cli + Chrome ────────────────────────────────
 # @playwright/cli  — token-efficient browser CLI used by research agents
 #                    (playwright-cli -s=research open/snapshot/click/type …)
-# playwright       — provides `playwright install` to download managed Chromium
+# playwright-cli defaults to Chrome at /opt/google/chrome/chrome. Install it directly.
-# Chromium is installed to /root/.cache/ms-playwright/ (default for root user)
+RUN npm install -g @playwright/cli \
-RUN npm install -g @playwright/cli playwright \
+    && curl -fsSL https://dl.google.com/linux/direct/google-chrome-stable_current_amd64.deb -o /tmp/chrome.deb \
-    && playwright install chromium
+    && apt-get update && apt-get install -y /tmp/chrome.deb \
+    && rm -f /tmp/chrome.deb && rm -rf /var/lib/apt/lists/*
 
 # ── Python tooling ────────────────────────────────────────────────────────────
 RUN pip install uv

bundle/agents/researcher.md

@@ -97,11 +97,24 @@ extract current data, and synthesize findings into clear, actionable results. Th
 browser runs persistently — tabs stay open between commands, and the user can watch
 and take over at any time via the noVNC view.
 
+## IMPORTANT: Pre-installed Environment
+
+`playwright-cli` is **already installed** in this container. Do NOT install anything.
+Do NOT run `npm install`, `pip install`, or any package manager. Just use the commands
+below directly via the bash tool.
+
 ## Browser Control (playwright-cli)
 
-You control the browser using `playwright-cli` with a persistent named session
+You control the browser using `playwright-cli` via the bash tool. All commands use
-(`-s=research`). All commands operate on the same browser session, preserving tabs
+a persistent named session (`-s=research`), preserving tabs and state across the
-and state across the entire conversation.
+entire conversation.
+
+**How to invoke:** Every playwright-cli command runs through the bash tool:
+```
+bash("playwright-cli -s=research open https://example.com")
+bash("playwright-cli -s=research snapshot")
+bash("playwright-cli -s=research click @e5")
+```
 
 ### Commands
 
@@ -109,9 +122,9 @@ and state across the entire conversation.
 |--------|---------|
 | Open a URL | `playwright-cli -s=research open https://example.com` |
 | Take a snapshot | `playwright-cli -s=research snapshot` |
-| Compact snapshot | `playwright-cli -s=research snapshot -ic` |
 | Click an element | `playwright-cli -s=research click @e5` |
-| Type text | `playwright-cli -s=research type @e3 "search query"` |
+| Type text | `playwright-cli -s=research type "search query"` |
+| Fill a field | `playwright-cli -s=research fill @e3 "search query"` |
 | Press a key | `playwright-cli -s=research press Enter` |
 | Scroll down | `playwright-cli -s=research scroll down` |
 | Go back | `playwright-cli -s=research back` |

bundle/bundle.md

@@ -10,8 +10,8 @@ providers:
     config:
       default_model: claude-sonnet-4-20250514
 
-orchestrator:
+session:
-  module: loop-streaming
+  orchestrator:
     config:
       extended_thinking: false
 

bundle/context/researcher-instructions.md

@@ -3,22 +3,56 @@
 This session runs inside the **Research Workbench**. You have access to the
 following capabilities:
 
-## Available Capabilities
+## CRITICAL: Pre-installed Tools
 
-**Browser** — Live Chromium browser controlled via `playwright-cli -s=research`.
+All tools are **already installed**. Do NOT run npm install, pip install, or any
-The browser is persistent across commands and visible to the user via VNC (noVNC
+package manager. Just use the commands directly via the bash tool.
-panel). The user can take over at any time.
 
-**Artifacts** — Persistent markdown documents written to `artifacts/<session_id>/`.
+## Browser Control
+
+Live Chromium browser controlled via `playwright-cli -s=research` through the
+bash tool. The browser is persistent across commands and visible to the user via
+the noVNC panel. The user can take over at any time.
+
+**Quick start — open a site and see what's on it:**
+```
+bash("playwright-cli -s=research open https://google.com")
+bash("playwright-cli -s=research snapshot")
+```
+
+**Click an element from the snapshot:**
+```
+bash("playwright-cli -s=research click @e5")
+bash("playwright-cli -s=research snapshot")
+```
+
+**Type into a field and submit:**
+```
+bash("playwright-cli -s=research fill @e3 \"search query\"")
+bash("playwright-cli -s=research press Enter")
+bash("playwright-cli -s=research snapshot")
+```
+
+Always snapshot after navigation to get fresh element refs.
+
+## Artifacts
+
+Persistent markdown documents written to `artifacts/<session_id>/`.
 They appear in the right panel and survive across sessions.
 
-**A2UI** — Inline interactive controls (filter chips, tables, card grids, forms)
+## A2UI
+
+Inline interactive controls (filter chips, tables, card grids, forms)
 rendered directly in chat via `a2ui` code blocks.
 
-**Inline Visuals** — Rich visual output (charts, maps, comparison tables with color
+## Inline Visuals
+
+Rich visual output (charts, maps, comparison tables with color
 coding) rendered directly in chat via `visual` code blocks.
 
-## Full Command Reference
+## Research Workflow
 
-For complete command syntax, workflow guidance, and best practices, see the
+1. Understand the request — clarify scope and constraints
-researcher agent system prompt.
+2. Plan — identify 2-4 sources to consult
+3. Browse and extract — use playwright-cli via bash to navigate sites and extract data
+4. Present — use A2UI, visuals, or artifacts to deliver results