aped-lead

Approve, escalate, and dispatch sprint Story check-ins across parallel worktrees.

2|Updated Mar 12, 2026
One-click install
npx skills add https://github.com/yabafre/aped-claude --skill aped-lead
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: aped-lead
Source: https://github.com/yabafre/aped-claude/tree/main/src/templates/skills/aped-lead
Command: npx skills add https://github.com/yabafre/aped-claude --skill aped-lead

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It resolves pending Story check-ins across parallel sprint worktrees by running deterministic approval checks, approving what is safe, and escalating what needs human input—so your sprint pipeline advances without inconsistent state.

Core Features & Use Cases

  • Parallel-sprint coordination: Reads the sprint graph and processes queued check-ins from Story Leaders at transition points (story-ready, dev-done, review-done), then pushes the next command back to the correct worktree.
  • Guardrailed, programmatic auto-approvals: For DONE-only items, runs scripted checks via check-auto-approve.sh and uses its exit codes as the source of truth.
  • Human-in-the-loop escalation: Treats dev-blocked as always ESCALATE, and for concerns or missing context (DONE_WITH_CONCERNS / NEEDS_CONTEXT) asks the user to decide or provides the blocking reason verbatim.
  • Load-bearing review-done merge handling: Merges approved story PRs into the sprint umbrella in the correct order (capture/merge+poll/record merged flag/mark done/teardown), then instructs the user to run aped-ship later for the umbrella release PR.
  • Safety and audit consistency: Enforces root-only execution, validates state integrity, and ensures every mutation goes through checkin.sh or sync-state.sh to preserve the audit trail.

Quick Start

Run aped-lead from the main project root when Story Leaders have posted pending check-ins so it can approve safe items and escalate the rest for your confirmation.

Frequently Asked Questions about aped-lead

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

FAQPage Schema
How do I coordinate sprint approvals across parallel worktrees?

Sprint approval coordination across parallel worktrees is handled by reading the sprint graph, processing queued Story check-ins at transition points, and pushing the next command to the correct worktree.

Can I automate pull request merging into a sprint umbrella?

Yes, PR merging into a sprint umbrella is automated by capturing, merging, polling, recording the merged flag, marking the story done, and tearing down the worktree in the correct ordered sequence.

How does human-in-the-loop escalation work for dev-blocked worktrees?

Human-in-the-loop escalation treats dev-blocked items as always requiring escalation, while DONE_WITH_CONCERNS or NEEDS_CONTEXT statuses prompt the user to decide or provide the exact blocking reason.

What is the best way to enforce deterministic auto-approvals for sprint check-ins?

Deterministic auto-approvals are enforced by running scripted checks via check-auto-approve.sh and using its exit codes as the source of truth for safely approving DONE-only items.

Do I need root execution to run state validation for sprint worktrees?

Yes, root-only execution is required to run state validation, ensuring every state mutation goes through checkin.sh or sync-state.sh to preserve the audit trail.

When should I escalate a review-done check-in instead of auto-approving?

Review-done check-ins should be escalated instead of auto-approved when agent_status indicates missing context, confidence-based routing is uncertain, or concerns are flagged that require human verification before merging.