ether-orchestrator

Dispatches pending ether-pro jobs to parallel subagents grouped by target repository.

Updated Jul 27, 2026
One-click install
npx skills add https://github.com/arayaroma/ether --skill ether-orchestrator-arayaroma
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ether-orchestrator
Source: https://github.com/arayaroma/ether/tree/main/skills/ether-orchestrator
Command: npx skills add https://github.com/arayaroma/ether --skill ether-orchestrator-arayaroma

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Draining multiple pending ether-pro jobs serially in one session forces unrelated repositories to wait on each other and bloats a single context window. This Skill fans independent work out to fresh parallel subagents so jobs across different target repos run concurrently. ## Core Features & Use Cases - Queue draining and grouping: Runs ether job pending across all job types, then partitions jobs by their target repo so only independent targets are parallelized. - Parallel subagent dispatch: Issues one Agent tool call per target group in the same message, building each subagent prompt from the shared role files at skills/_shared/roles/<job-type>.prompt. - Confirmation gating: Surfaces any should-confirm pauses from subagents to the user instead of letting parallel execution become unattended, and summarizes one line per target. - Use Case: You have five pending jobs — two repo reviews on repo A, a PR author job on repo B, and two PR reviews on repo C. The Skill dispatches three subagents in parallel, one per repo, and reports outcomes, tiers, and artifacts per target. ## Quick Start Run all pending ether jobs across my repos in parallel and summarize the results per target.

Frequently Asked Questions about ether-orchestrator

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run all pending ether-pro jobs in parallel?▼

Invoke the ether-orchestrator skill when two or more pending jobs span different target repos. It runs `ether job pending`, groups jobs by target, and dispatches one subagent per group in parallel via the Agent tool.

When should I use the orchestrator instead of running a job inline?▼

Use it only when pending jobs cover at least two independent target repos. A single job, or multiple jobs on the same target, should run inline through the job type's own skill since orchestration adds subagent context overhead with no parallelism benefit.

Can two jobs on the same repository run in parallel?▼

No. Jobs sharing a target are never parallelized because they can race on the same git branch or worktree, and a pr-review job could review a PR mid-write by a pr-author job. Same-target jobs run sequentially in one subagent.

Does parallel dispatch bypass confirmation prompts for write actions?▼

No. Each job's should-confirm tier gate still applies inside subagents, since tiering is per job type and action type in jobs.db. The orchestrator surfaces any pending confirmation to the user before treating a group as done.

What happens if a dispatched subagent fails to report back?▼

A missing or malformed subagent result is treated as outcome failure for that job's record-run, not silence. The orchestrator calls out any subagent that failed to report cleanly in its per-target summary.