leek/agent-skills / skills/implement-spec
implement-spec
VerifiedBuild a whole spec fast: parallel implementer subagents work the ticket frontier, each in its own worktree, merging onto one integration branch; then one review, one verification pass, and one PR or a stack of PRs.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 28.9 KB |
| agents/openai.yaml | config | 228 B |
| references/laravel.md | markdown | 691 B |
| references/leftovers.md | markdown | 846 B |
| references/resume-and-gates.md | markdown | 3.5 KB |
| scripts/screenshots-report.shexecutable | script | 3.5 KB |
| scripts/spec-run-state.shexecutable | script | 5.6 KB |
| scripts/test-screenshots-report.shexecutable | script | 3.4 KB |
| scripts/test-spec-run-state.shexecutable | script | 7.2 KB |
| scripts/test-worktree-idle.shexecutable | script | 2.2 KB |
| scripts/worktree-idle.shexecutable | script | 2.1 KB |
scripts/worktree-idle.sh
View on GitHub#!/bin/bash
# Block until a worktree has gone N minutes with no file change, then print one
# JSON line and exit 0. The stall alarm for a lone implementer: run it under a
# Monitor (or as a background task) and its exit is the wake-up.
#
# Usage: worktree-idle.sh <worktree> [minutes] (default 20)
# Prints {"worktree":"…","status":"idle","idle_seconds":N} or, when the worktree
# is removed mid-watch, {"worktree":"…","status":"gone"}. Exits 1 on bad input.
#
# Portable (macOS and Linux): it never reads file times. Each poll asks `find`
# for files newer than a marker touched at the previous poll, and keeps the
# last-change time as a shell clock. .git, vendor and node_modules are skipped.
# Env (tests): WORKTREE_IDLE_LIMIT_SECONDS overrides minutes,
# WORKTREE_IDLE_POLL_SECONDS sets the poll interval (default 60).
set -e
dir=${1:?usage: worktree-idle.sh <worktree> [minutes]}
minutes=${2:-20}
case "$minutes" in ''|*[!0-9]*) echo "worktree-idle: minutes must be a whole number: $minutes" >&2; exit 1 ;; esac
[ -d "$dir" ] || { echo "worktree-idle: not a directory: $dir" >&2; exit 1; }
dir=$(cd "$dir" && pwd -P)
limit=${WORKTREE_IDLE_LIMIT_SECONDS:-$((minutes * 60))}
poll=${WORKTREE_IDLE_POLL_SECONDS:-60}
tmp=$(mktemp -d)
trap 'rm -rf "$tmp"' EXIT
marker="$tmp/marker"
touch "$marker"
last_change=$(date +%s)
changed_since() { # changed_since <marker>: true when any watched file is newer
find "$dir" \( -name .git -o -name vendor -o -name node_modules \) -prune \
-o -newer "$1" -print 2>/dev/null | grep -q .
}
echo "worktree-idle: watching $dir; alarm after ${limit}s with no file change" >&2
while :; do
sleep "$poll"
if [ ! -d "$dir" ]; then
jq -cn --arg w "$dir" '{worktree:$w,status:"gone"}'
exit 0
fi
# Touch the next marker before the scan so a write during the scan is not lost.
touch "$tmp/next"
now=$(date +%s)
if changed_since "$marker"; then
last_change=$now
fi
mv "$tmp/next" "$marker"
idle=$((now - last_change))
if [ "$idle" -ge "$limit" ]; then
jq -cn --arg w "$dir" --argjson s "$idle" '{worktree:$w,status:"idle",idle_seconds:$s}'
exit 0
fi
done
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 469k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |