loopx

Autonomously triage, implement, and ship every pending item in a .pm workstream.

275|32|Updated Oct 2, 2020
One-click install
npx skills add https://github.com/bex-co/beancount-io --skill loopx-bex-co
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: loopx
Source: https://github.com/bex-co/beancount-io/tree/main/.agents/skills/loopx
Command: npx skills add https://github.com/bex-co/beancount-io --skill loopx-bex-co

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? Project boards accumulate stale milestones, tasks, and inbox notes that nobody systematically works through. This Skill drains an entire .pm workstream autonomously, so backlog items get implemented, closed, blocked, or deleted based on evidence instead of lingering unchecked. ## Core Features & Use Cases - Autonomous backlog draining: Repeatedly picks the next actionable pending milestone, task, or inbox note in a workstream and processes it until nothing actionable remains. - Evidence-based triage: Every item is verified against the actual repository state (git log, grep, acceptance criteria) before being worked on, closed as already done, blocked with an unblock condition, or deleted as invalid. - Integrated shipping: Each item's outcome is committed and pushed individually via the /ship skill, with all .pm/ board writes going through the /pm skill. - Use Case: Run $loopx w1 to have the agent work through every pending item in workstream w1 overnight — implementing features, closing stale tasks, and blocking items that need your input, with a full report at the end. ## Quick Start Ask the agent to run $loopx w1 to drain every pending milestone, task, and inbox note in workstream w1 until nothing actionable remains.

Frequently Asked Questions about loopx

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

FAQPage Schema
How do I drain an entire project backlog autonomously with an AI agent?

Invoke $loopx with a workstream identifier, such as $loopx w1. The agent repeatedly picks the next actionable pending milestone, task, or inbox note, triages it against the repository state, implements or resolves it, ships the change, and continues until nothing actionable remains.

How does loopx decide whether to implement, close, block, or delete a task?

Every item goes through evidence-based triage before any code is written. The agent checks git history, greps for referenced commands and paths, and runs acceptance criteria. Items are worked on if still needed, closed if already satisfied, blocked with an unblock condition if stuck, or deleted only with citable evidence of invalidity.

What is the difference between loopx and a timed loop command?

loopx is sequential, not interval-based — it works through a workstream's backlog item by item until done. For a timed poll that re-checks on an interval, use the separate /loop skill instead.

What happens when loopx hits a task it cannot complete?

Unresolvable items are moved to blocked/ via /pm block with an exact unblock condition and owner. Partial work is preserved in an isolated worktree or stash, and the loop continues with the next independent actionable item rather than stopping.

When should I not use loopx?

Do not use loopx for a single task, ordinary code edits, or timed polling. It is designed only for draining a whole workstream's backlog and requires running on the main branch with a clean understanding of uncommitted changes.