QA: Sub-Agents — AWS Strands

Prerequisites

• Demo is deployed and accessible at /demos/subagents on the dashboard host
• Agent backend is healthy (/api/health); OPENAI_API_KEY is set; the Strands agent server (agent_server.py) is reachable at AGENT_URL

Test Steps

1. Basic Functionality

• Navigate to /demos/subagents; verify the page renders within 3s with a wide delegation log on the left and a CopilotChat pane on the right
• Verify data-testid="delegation-log" is visible with heading "Sub-agent delegations"
• Verify the empty state reads "Ask the supervisor to complete a task. Every sub-agent it calls will appear here."
• Verify data-testid="delegation-count" reads "0 calls"
• Verify the chat input placeholder is "Give the supervisor a task..."
• Verify all 3 suggestion pills are visible: "Write a blog post", "Explain a topic", "Summarize a topic"

2. Feature-Specific Checks

Supervisor delegates to sub-agents (research → write → critique)

• Click the "Write a blog post" suggestion (sends a research/write/critique sequence prompt)
• Within 5s verify data-testid="supervisor-running" appears next to the heading with the pulsing indicator
• Within 60s verify data-testid="delegation-count" reads at least "3 calls" and at least 3 data-testid="delegation-entry" rows are rendered
• Verify the first entry has the 🔎 Research badge, a non-empty Task: line, and a result body containing 3-5 bullet points
• Verify a subsequent entry has the ✍️ Writing badge with a polished paragraph in the result body
• Verify a subsequent entry has the 🧐 Critique badge with 2-3 actionable critiques
• Verify each entry shows status completed (green) once the sub-agent has returned
• Once the supervisor returns a final assistant text message, verify supervisor-running disappears

Live updates during the run

• Click the "Explain a topic" suggestion
• Verify entries appear ONE-AT-A-TIME (the count climbs from 0 → 1 → 2 → 3 over multiple seconds rather than all appearing at once) — this confirms each sub-agent's state_from_result hook emits an independent StateSnapshotEvent
• Verify entry numbers (#1, #2, #3) are sequential and stable

Multi-turn persistence

• After the first run completes, send "Now do the same for solar panels." — verify the delegation log GROWS (existing rows kept; new rows appended); the count should continue from where it left off (e.g. "3 calls" → "6 calls")

3. Error Handling

• Send an empty message — verify it is a no-op
• Send "Hello" (no task to delegate); verify the supervisor responds with a short text reply and NO new delegation entries are added
• Verify DevTools -> Console shows no uncaught errors during any flow above

Expected Results

• Page loads within 3 seconds
• First delegation entry appears within 15s of submitting a task; full research → write → critique loop completes within 60s
• Each sub-agent invocation produces exactly one delegation entry with a non-empty task and result
• Supervisor's final summary references the work that was delegated
• No UI layout breaks, no uncaught console errors