Automated Bugzilla bug workflow: scrape, analyze, plan, execute, verify. Orchestrates a multi-agent pipeline from bug report to verified fix.
npx claudepluginhub konstantilieris/bugfow_pluginExplore the codebase, analyze the bug or feature request, and determine FE/BE/MIXED/FEATURE verdict.
Authenticate to a web application for Playwright testing. Reads all credentials, URLs, and selectors from config. Supports SSL bypass for self-signed certificates.
Create a comprehensive backend analysis document for bugs with BE or MIXED verdict.
Handle Bugzilla authentication reliably with explicit verification.
Post a completion comment to Bugzilla after a bug fix is applied and verified.
Scrape a Bugzilla bug page and extract structured data using Playwright MCP.
Create the bug dossier file and state file from Bugzilla data.
Coordinates the execution and verification phase with checkpoint rollback and cleanup orchestration.
Apply the approved implementation plan by making code changes with transactional execution.
Generate a minimal-risk frontend implementation plan.
Design and implement new features using Claude skills.
Coordinates the initial ingest phase of the bug workflow including Scout, Dossier, and Branch creation.
Extracts reusable patterns from completed bug workflows for cross-workflow learning.
Coordinates the planning and validation phase with revision loop and anti-pattern tracking.
Validate implementation plan before user review ensuring minimum impact and best practices.
Validate all prerequisites before execution begins to prevent wasted work.
Send workflow notifications to a Slack channel at key workflow events.
Generate a structured testSpec JSON for automated Playwright verification.
Execute structured test specifications mechanically and collect evidence from console logs.
Execute approved bug fix plan after user approves. Use after /bug-start or /bug-plan creates a plan. Requires state PLAN_FE, VALIDATE, FEATURE_DESIGN, or AWAITING_APPROVAL.
Generate minimal-risk FE implementation plan for a bug. Use when bug workflow is started, verdict is FE/MIXED, and implementation plan is needed. Requires existing state file.
Reset bug workflow state to allow re-running
Resume interrupted workflow from last checkpoint
Interactive setup wizard for the Bugflow plugin. Generates project-local configuration files.
Start bug workflow when beginning work on a Bugzilla ticket. Scrapes Bugzilla, creates dossier, analyzes code, generates plan. Use with a bug ID to begin automated workflow.
Show current workflow state for a bug
Return FE/BE/Mixed verdict for an analyzed bug with reasoning
Run verification tests to confirm bug fix works
Automated Bugzilla bug workflow that orchestrates a multi-agent pipeline: scrape bug data, analyze root cause, plan the fix, execute code changes, and verify with browser tests.
.mcp.json)# Add the marketplace
/plugin marketplace add Konstantilieris/Bugfow_Plugin
# Install the plugin
/plugin install bugflow@Konstantilieris
Or for local development:
claude --plugin-dir /path/to/bugflow-plugin
Run the setup wizard to configure the plugin for your project:
/bugflow:bug-setup
This will ask for:
Configuration is saved to your project's .claude/bugflow/config/ directory (auto-added to .gitignore).
| Command | Description |
|---|---|
/bugflow:bug-start <id> | Start full pipeline: scrape, analyze, plan (stops for approval) |
/bugflow:bug-accept <id> | Execute approved plan, verify, notify |
/bugflow:bug-plan <id> | Re-generate implementation plan |
/bugflow:bug-verify <id> | Re-run verification tests |
/bugflow:bug-status <id> | Check current workflow state |
/bugflow:bug-reset <id> | Reset workflow state for a bug |
/bugflow:bug-resume <id> | Resume interrupted workflow from last checkpoint |
/bugflow:bug-verdict <id> | Get FE/BE/MIXED/FEATURE verdict |
/bug-start <id>
-> Scout: Scrape Bugzilla
-> Dossier: Create analysis doc + git branch
-> Analysis: Determine verdict (FE/BE/MIXED/FEATURE)
-> Plan: Generate implementation plan
-> Validate: Score plan quality (loop if < 70)
-> [STOP] Display plan for user approval
/bug-accept <id>
-> Pre-flight: Validate prerequisites
-> Execute: Apply code changes
-> Verify: Run browser tests
-> Notify: Slack + Bugzilla comment
-> Extract: Learn patterns for future bugs
| Verdict | Meaning | Action |
|---|---|---|
| FE | Frontend bug fix | FE Implementer Planner agent |
| BE | Backend bug | Creates handoff document |
| MIXED | Both FE and BE | FE plan + BE handoff doc |
| FEATURE | New feature request | Feature Designer agent |
The plugin ships with 19 specialized agents that handle different stages of the workflow. See infrastructure/ARCHITECTURE.md for the full agent reference.
After running /bugflow:bug-setup, two config files are generated:
environment.json — URLs, Slack channel, paths, workflow settingsplaywright.json — Browser config, credentials, selectorsThese live in your project's .claude/bugflow/config/ and should NOT be committed to git (they contain credentials).
Edit .claude/bugflow/config/playwright.json to add selectors for your app:
{
"selectors": {
"loginForm": {
"username": "#your-username-field",
"password": "#your-password-field",
"submitButton": "#your-login-button"
},
"app": {
"mainContent": ".your-main-content",
"navigation": ".your-nav"
}
}
}
Edit environment.json to adjust:
workflow.maxPlanRevisions — how many times to retry plan generation (default: 3)workflow.maxVerifyRetries — how many times to retry verification (default: 2)workflow.lockExpiryHours — how long a workflow lock lasts (default: 4)MIT
Slash commands for sisyphus multi-agent orchestration
Bypasses permissions
Runs without the normal permission approval flow
Uses power tools
Uses Bash, Write, or Edit tools
Share bugs, ideas, or general feedback.
Streamlines bug fixing by creating a GitHub issue first, then a feature branch for implementing and thoroughly testing the solution before merging.
Find and fix a bug. Default is the careful path (reproduce, investigate, test). Add `--fast` for emergency hotfix mode.
Streamlines bug fixing by creating a GitHub issue first, then a feature branch for implementing and thoroughly testing the solution before merging.
127-agent automated development system with Agent Teams, quality gates, Bug Council diagnostics, and autonomous execution
Reliable automation, in-depth debugging, and performance analysis in Chrome using Chrome DevTools and Puppeteer
Share bugs, ideas, or general feedback.