Whilly Orchestrator

Task orchestrator that runs coding-agent CLIs on a JSON plan — loopable, resumable, budget-capped, and happy to hand tasks back to a human when it hits a wall.

Getting Started View on GitHub

What whilly does

  • Pulls tasks from a GitHub repo (--from-github, --from-issue), a Projects v2 board (--from-project), or a Jira project (--from-jira).
  • Runs agents — Claude CLI by default, OpenCode, or claude_handoff (file-based RPC to you).
  • Tracks lifecycle on your Projects v2 board and your Jira ticket in real time: Todo → In Progress → In Review → Done, plus On Hold and Human Loop for anything that needs a human.
  • Stays safe — hard budget cap, wall-clock timeout, automatic task resume, idempotent plan files.
  • Works everywhere — pure Python 3.10+, stdlib-only for Jira, tested on Linux / macOS / Windows.

One-liner demo 

pipx install whilly-orchestrator
whilly --config path            # where to drop your config
whilly --from-issue you/repo/42 --go

That fetches issue 42, generates a one-task plan, spawns an agent, and exits 0 when the agent reports complete.

Page When to read
Getting Started First time here — eight practical walkthroughs
Full Usage Reference Every CLI flag, env var, and config field
GitHub Integration Guide Setting up Projects v2 + board sync
Interfaces & Tasks Module contracts + the JSON plan schema
Architecture Decisions Why things are the way they are (if published)

Under the hood 

CLI  ──▶  TaskManager  ──▶  AgentBackend  ──▶  Claude / OpenCode / claude_handoff
                ↓
   on_status_change hook  ──▶  Projects v2 (GraphQL)  +  Jira (REST)
                ↓
         Dashboard, Reporter, Budget + Resource guards

Full module map lives in Whilly-Interfaces-and-Tasks.

Current status

  • 643 tests, cross-OS CI (Linux / macOS / Windows) on every PR.
  • Layered config — whilly.toml + OS keyring, migrates from legacy .env with one command.
  • Latest release · Open issues · Changelog