Workflows & Automation
Turn your kanban board into an automated development pipeline. Queue tasks for AI agents, review pull requests, watch your desktop's processes, and let AI resolve merge conflicts — control it all from your phone or your desktop, while every run executes on your paired desktop.
Kanban Boards
Connect a board to Jira through OAuth, or skip Jira entirely and run a board that lives only in Hermes. Either way, you get the same kanban experience for planning and tracking work.
- Jira integration via OAuth 3LO — connect boards from any Jira project
- Local boards that need no Jira account, with offline card creation
- Search cards, filter columns, and switch between multiple boards
- Cards sync bidirectionally with Jira the moment you're connected
Manage boards from your phone or your desktop — the same board updates on both.
What you need
A Jira account with OAuth access for Jira-connected boards — or nothing at all for local boards.
AI Conveyor Pipeline
The conveyor turns a kanban board into an automated development pipeline. A card moves through a sequence of AI-run stages, from first read to a finished pull request.
Claude, Codex, or Gemini drives each stage — you choose which, per board or per card. Plan Validation sends a rejected plan back to Planning. AI Review sends a failing audit back to Implementation. A failed test sends the card back to Implementation too.
Large stories decompose into parallel subtasks, each working in its own git worktree, and human gates pause the pipeline at key points until you approve. When a card clears review, the conveyor opens a pull request for it automatically.
- Human gates pause the pipeline for your approval before it proceeds
- Story decomposition fans large stories out into parallel subtasks, each in its own git worktree
- Automatic PR creation once implementation passes review
- Per-card controls: retry, stop, resume, force advance, cancel, or switch CLI
What you need
At least one AI CLI installed — Claude Code, Codex, or Gemini CLI — a git repository, and a Jira or local board with the conveyor enabled.
Task Queue
Queue a task for Claude, Codex, or Gemini without touching a kanban board. Start it from your phone, watch the output stream live, and review the result when it's done.
- Monitor live output streaming as the task runs
- Review results and iterate without switching tools
- Tasks execute on your paired desktop; you control them from mobile
- Organize tasks per project
What you need
At least one AI CLI installed on your desktop.
Pull Requests
Browse every pull request for a project from your phone. Open one to see the overview, changed files, full diff, comments, and check runs.
- Read bot review comments from Copilot, CodeRabbit, and SonarCloud
- Submit your own comments directly from mobile
- Trigger an AI code review — a full worktree checkout reviewed with your configured CLI, model, and skill
- Create a PR directly from a finished task or a kanban card
What you need
A GitHub repository connected through the Hermes GitHub integration.
Process Monitor
Check on your desktop's health from anywhere. The process monitor gives you a live system dashboard and control over what's running, right from your phone.
System Dashboard
CPU usage, memory used and total, and load average at a glance.
Process List
PID, name, command, CPU%, and memory for every running process.
Sort and Filter
Sort by CPU, memory, or name, and filter the list by text.
Kill Processes
End a process with SIGTERM or SIGKILL — a confirmation dialog stands between you and the kill.
Log Tailing
Stream system.log, syslog, or a custom log file path in real time.
What you need
A paired desktop connection.
AI Merge Resolution
When a worktree merge hits a conflict, the conveyor doesn't stop and wait — it resolves it. Your configured AI CLI reads the conflicting files and rewrites them.
- Validates the resolution by checking for remaining conflict markers
- Stages resolved files with
git add - Returns a clear outcome: resolved, unresolved (with the remaining files listed), stale, or failed
- Runs automatically during multi-worktree story decomposition, as part of the conveyor pipeline
What you need
An AI CLI installed and an active merge conflict in a git worktree.