Objective
Audit http://localhost:3001/demo-target for conversion + accessibility issues. Use browser_qa_audit.
Agents done
0/6
Progress
0%
Total tokens
0
Total latency
0.00s
Created
Queued
Running
Waiting approval
Receipt
A mission is one run of a crew. Agents execute in order; prompts, raw outputs, parsed JSON, tools, memory, approvals, and receipts are persisted.
What this page is
The black-box recorder for one mission.
Each agent step
Saved with prompt, raw model output, latency, and token counts.
Raw model output
Stored verbatim alongside the parsed JSON, so you can replay or audit.
Parsed JSON
Validated against the agent's Zod schema; failures self-correct or fail loudly.
Tool calls
Every browser_qa_audit, web_snapshot, etc. is recorded with input + output.
Approvals
Risky outputs (drafts, trades, exports) wait for your decision before they count.
Receipt
A WorkflowReceipt with an integrity hash captures the whole run for the trust ledger.
This page auto-refreshes while the mission is queued or running.
Crew
Explorer Agentpending
Maps pages and key flows via real browser audit.
Flow Tester Agentpending
Tests signup/contact/checkout-style flows.
Accessibility Agentpending
Checks labels, contrast notes, keyboard risks.
Copy Agentpending
Flags confusing or weak copy.
Bug Reporter Agentpending
Creates recommended fixes for each issue.
Report Agentpending
Drafts a client-ready QA report summary.
Run log
No agent steps yet. Click Run mission to start the crew.
Workflow receipt
No receipt yet for this mission. Generate one to capture an immutable record of agents, tools, approvals, memory used, and risk.
Audit log
May 1, 01:31 AM
user
mission.created
Created mission "Itil"