How it works

Bugzy participates in your software development lifecycle the way a QA teammate would — reacting to changes, proposing actions, executing tests, and reporting back. This page explains the interaction model.

Triggers and tasks

Everything starts with a trigger. A trigger runs a predefined or preconfigured task:

Trigger (manual via UI, Slack/Teams message, GitHub event, deployment, schedule)
  → Task executes (tests, triage, verification, exploration)
    → Results reported back (Slack, Teams, PR comment, dashboard)
      → Team provides feedback (dispute findings, ask questions)
        → Bugzy learns and improves

Some triggers execute automatically — cron schedules, deployment hooks, and GitHub webhooks run their configured task without manual confirmation. Conversational triggers like Slack or Teams messages go through the inbox agent, giving your team a chance to review before Bugzy acts.

Tasks

Bugzy uses a task-based architecture. Each task is a step-by-step workflow that composes subagents to accomplish a goal.

Task	What Bugzy does
`onboard-testing`	End-to-end onboarding: explore app, plan, generate tests, execute, fix, report
`generate-test-plan`	Creates a structured test plan from product description + browser exploration
`generate-test-cases`	Converts test plan into manual test docs + automated Playwright scripts
`run-tests`	Executes tests, triages, auto-fixes, files bugs, notifies team
`verify-changes`	Runs relevant tests after a code change (PR, deployment, ticket)
`triage-results`	Classifies failures as product bugs vs test issues
`explore-test-codebase`	Analyzes an existing test repo to understand structure and patterns
`handle-message`	Responds to team messages from Slack or Teams
`process-event`	Handles webhook events (GitHub, Jira, meeting transcripts)

Subagents

Subagents are specialized AI capabilities that tasks compose together. You configure which subagents are active and which integrations they use during project setup.

Subagent	Role	Always active
Browser Automation	Navigates pages, interacts with UI via Playwright	Yes
Test Engineer	Generates Playwright scripts, debugs failures, fixes flaky tests	Yes
Test Case Manager	Manages test cases via filesystem or external TMS (Zephyr Scale)	Yes
Team Communicator	Posts results to Slack, Teams, or email	Yes
Issue Tracker	Files bugs in Jira, Azure DevOps, Asana, or Linear	When connected
Documentation Researcher	Searches Notion, Confluence, or Jira wikis for product context	When connected
Source Control	Reads PR/commit history, posts PR comments, creates check runs	When connected

Getting started

Use Cases

Capabilities

Integrations

Platform

Security

How it works

Triggers and tasks

Tasks

Subagents

Deep dive

Execution environment

Extend test coverage

​Triggers and tasks

​Tasks

​Subagents

​Deep dive

Execution environment

Extend test coverage

Triggers and tasks

Tasks

Subagents

Deep dive