aops-loop-interactive

Guides operators through the aops-cli loop lifecycle with gated supervised runs and worktree fan-out.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/eeemzs/aops --skill aops-loop-interactive-eeemzs
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: aops-loop-interactive
Source: https://github.com/eeemzs/aops/tree/main/assets/skills/aops-loop-interactive
Command: npx skills add https://github.com/eeemzs/aops --skill aops-loop-interactive-eeemzs

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Running multi-agent implementation loops without a clear playbook leads to ungated real runs, missed review steps, and confusion over which system owns planning, decisions, and durable state. This Skill gives operators a question-first, gate-driven playbook for the entire aops-cli loop lifecycle. ## Core Features & Use Cases - Readiness and pack gating: Runs loop plan dry-run readiness checks and loop pack previews before any real execution, surfacing failed gates and fixes. - Double-gated supervised runs: Enforces the mandatory --run-agents plus --supervised-run-approved pair before starting or resuming child-agent sessions, including resume-on-trigger via loop watch. - Parallel worktree fan-out (v2): Orchestrates the gated parallel-plan → worktree-init → maker-run → independent review → slice-merge → cleanup pipeline with one maker per slice and ff-only merges. - Use Case: An operator wants to fan a sprint out across three isolated git worktrees with Codex as maker and an independent reviewer per slice. The Skill walks them through readiness checks, per-slice PM review-requests, supervised maker runs, and operator-approved sequential merges. ## Quick Start Ask the agent to use the aops-loop-interactive skill to plan and start a supervised aops-cli loop for your task and sprint, beginning with a dry-run readiness check.

Frequently Asked Questions about aops-loop-interactive

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

FAQPage Schema
How do I start a supervised aops-cli loop run?▼

Run `aops-cli loop plan` first as a dry-run readiness check, then `loop pack` to preview refs. A real run requires both `--run-agents` and `--supervised-run-approved` flags in the same explicit operator instruction; neither gate can be inferred from prior approvals.

How do I run parallel agents across git worktrees with aops-cli?▼

Use the v2 fan-out pipeline: `loop parallel-plan`, `loop worktree-init`, then one gated `loop maker-run` per slice. Each slice needs its own PM review-request, an independent reviewer, and an operator-approved ff-only `slice-merge` before cleanup.

Can aops-cli loop resume an existing agent session?▼

Yes. Start with `--session-persist` to capture a session id, then use `loop resume-agent --session-id <id>` with a new directive. `loop watch` performs a single poll and resumes on a qualifying trigger, requiring `--watch-ignore` to avoid self-triggering.

Does the aops-cli loop facade own durable run state?▼

No. The loop facade is local and stateless by default. Durable run records, worker leases, heartbeats, retries, and reconcile state belong to the hosted Runner, used only when `--hosted-runner-mode=require-hosted` is explicitly set.

Why does aops-cli loop start refuse to run agents?▼

The start command blocks when readiness gates fail or when one of the two real-run gates is missing. Check the `loop plan` JSON output for failed checks such as missing reviewer evidence, unbounded budget, or absent `--supervised-run-approved`.