merge-wave

A workflow for reviewing several GitHub pull requests (proposed code changes) after their test262 test runs finish. It compares each change by how many JavaScript tests pass or fail, then orders and processes the requests.

In plain words
What is it for?
Use it to collect test reports from multiple pull requests, rank them by their net change in passing tests, merge the best candidates, and close harmful ones.
Why use it?
It removes the need to inspect completed test results and decide the merge order by hand. It also identifies changes that improve results and those that make them worse.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/loopdive/js2/merge-wave
Any agent
npx skills add loopdive/js2 --skill merge-wave
Clone the repo
git clone --depth 1 https://github.com/loopdive/js2

Made for: Claude Code, Codex.

Per session 45 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,566 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00045 $0.01566
Opus 5 $0.00023 $0.00783
Sonnet 5 $0.00009 $0.00313
Haiku 4.5 $0.00005 $0.00157

Measured 2d ago against content hash 6caa2043fe8a, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

merge-wave scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/skills/merge-wave/SKILL.md · 146 lines

How it starts

The opening of the file, as written. The whole thing — 146 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Merge Wave Protocol

Triage every open PR that has finished sharded test262 CI, compute its real net delta, and execute a ranked merge order in one pass.

When to use

  • Multiple PRs ready to merge after a morning of dev work
  • After a sprint's worth of issues finished CI at roughly the same time
  • When you want to cross a pass-rate milestone in one batch

Step 1: Gather candidate PRs

gh pr list --limit 20 --json number,title,headRefName,mergeable,mergeStateStatus

Filter to PRs with mergeStateStatus: UNSTABLE | CLEAN | BEHIND — skip DIRTY (conflict, ping dev to refresh) and DRAFT.

Step 2: Download every PR's merged-report artifact

The sharded workflow uploads test262-merged-report per run. Find each PR's run ID via the merge shard reports check:

mkdir -p output/pr-artifacts
for pr in <LIST>; do
  run_id=$(gh pr view $pr --json statusCheckRollup --jq '.statusCheckRollup[] | select(.name=="merge shard reports") | .detailsUrl' | grep -oE 'runs/[0-9]+' | head -1 | cut -d/ -f2)
  mkdir -p output/pr-artifacts/pr-$pr
  gh run download $run_id -n test262-merged-report -D output/pr-artifacts/pr-$pr >/dev/null 2>&1
done

Artifact filenames: test262-report-merged.json, test262-regressions.txt, test262-results-merged.jsonl.

Step 3: Compute and rank net deltas

BASELINE=$(python3 -c "import json; d=json.load(open('benchmarks/results/test262-current.json')); print(d['summary']['pass'])")

for pr in <LIST>; do
  rpt=output/pr-artifacts/pr-$pr/test262-report-merged.json
  [ -f "$rpt" ] && python3 -c "
import json
d=json.load(open('$rpt'))
s=d['summary']
delta=s['pass']-$BASELINE
print(f'PR #$pr  pass={s[\"pass\"]}  delta={delta:+d}  ce={s[\"compile_error\"]}')
"
done | sort -t= -k3 -n -r

Step 4: Classify each PR

Thresholds (tune per context):

Delta Action
< −500 CATASTROPHIC → close immediately with diagnosis. Do NOT merge.
−500 to −20 Real regression → close, open follow-up issue, ask dev to investigate root cause
−20 to +20 MARGINAL → check if regressions are false positives; if yes merge, if no close
+20 to +100 GOOD → sample regressions, admin-merge if regressions are false positives or acceptable
+100+ STRONG → admin-merge, flag regressions in a follow-up issue

Read the full file on GitHub · 146 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 146 lines · 45 tokens per session scan A 6caa2043fe8a

Subscribe to this mod's changes

merge-wave is a skill published in the GitHub repository loopdive/js2 (59 stars, last pushed 2d ago), licensed Apache-2.0. It adds 45 tokens to every session and 1,566 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.