wave

A workflow for running several independent development tasks in parallel, with each task handled in its own worktree, a separate working copy of the repository.

In plain words
What is it for?
It is for executing a ready-made wave of independent plan items, reviewing the results, merging them in order, and creating one pull request.
Why use it?
It reduces waiting when a plan contains multiple tasks that do not depend on one another, while keeping their changes separate until they are merged.

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/progiri/craftlight/wave
Any agent
npx skills add progiri/craftlight --skill wave
Clone the repo
git clone --depth 1 https://github.com/progiri/craftlight

Made for: Claude Code, Codex.

Per session 197 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 5,441 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 $0.00197 $0.05441
Opus 5 $0.00098 $0.02721
Sonnet 5 $0.00039 $0.01088
Haiku 4.5 $0.00020 $0.00544

Measured 2d ago against content hash f1a63bc1b21f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

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.

plugins/craftlight/skills/wave/SKILL.md · 264 lines

How it starts

The opening of the file, as written. The whole thing — 264 lines — stays where its author put it; the contents beside it link to each section on GitHub.

wave — running one wave of a PLAN in parallel

Principle: wave sits BESIDE task, under plan. plan lays an initiative out into waves and stops at hand-off; task runs one leaf; wave runs one wave — N independent leaves in parallel, each a subagent in its own worktree, one human gate over the whole wave, one PR at the end. It is the execution layer plan deliberately refuses to be. Not for this skill: one task, however large → task; building or re-cutting the DAG → plan; a single leaf → a plain task call; review → code-review.

What you buy is wall-clock, not tokens: every executor reloads the project's context, so a wave's spend grows roughly ×N (measured: ~23–26 subagent tokens per token of the orchestrator's own growth) while the orchestrator's context stays cheap — it gives you no feedback at all about the money spent. Under 3 leaves, don't open a wave: the ceremony pays only on a wide one, where it costs the slowest leaf rather than the sum.

Two deliberate revisions, stated rather than slipped through. (1) M forbids execution subagents because each reloads the context — here that fan-out is the point, at the ×N price above. (2) A leaf does not enter task's router; it executes an already-approved spec, because task's gate waits for the user's next message and a subagent has none — so the classification task owns happens at the wave's gate, done by you (step 3). PR granularity is not a revision: plan already holds it — a leaf run solo lands its own PR, a leaf run inside a wave keeps its branch and the wave lands one PR (step 5).

Step 0. Orientation and resume

There's a CRAFT.md → read it first, then the initiative's plan at docs/crafts/<initiative>/PLAN.md. The PLAN is the wave's external memory: a wave interrupted mid-run resumes from the PLAN and from git, never from chat history — a compaction eats the chat and leaves the artifacts. The craftlight block upsert in the root CLAUDE.md (procedure — skills/task/templates/CLAUDE-block.md) rides with wave's first write to disk, whichever comes first — the PLAN append below or phase 1's specs — and never before one of them.

Read the full file on GitHub · 264 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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 · 264 lines · 197 tokens per session scan A f1a63bc1b21f

Subscribe to this mod's changes

wave is a skill published in the GitHub repository progiri/craftlight (5 stars, last pushed 7d ago), licensed MIT. It adds 197 tokens to every session and 5,441 once invoked, about $0.0010 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 2026-08-31.

Related

Other skills, from other repositories

systematic-debugging

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.

obra/superpowers · 21 tokens

chat-pet-sprite-creation

Use when creating or changing VS Code chat pet sprite art, sprite sheets, state animations, eye treatments, Stable/Insiders variants, or pet transitions under src/vs/workbench/contrib/chat/browser/widget/media/chatPet.

microsoft/vscode · 53 tokens

cpu-profile-analysis

Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…

microsoft/vscode · 71 tokens

babysit-pr

Babysit a GitHub pull request after creation by continuously polling review comments, CI checks/workflow runs, and mergeability state until the PR is merged/closed or user help is required. Diagnose failures, retry likely flaky failures up to 3 times, auto-fix/push branch-related issues when appropriate, and keep…

openai/codex · 114 tokens

imagegen

Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups, or transparent-background cutouts. Use when Codex should create a brand-new image, transform an existing image, or derive visual variants from references, and the output…

openai/codex · 113 tokens

agent-host-chat-contributions

Build and review cross-cutting agent-host chat behavior through lifecycle contributions. Use when adding turn lifecycle side effects, prompt or context injection, restored-history transformation, protocol-action observation, or when reviewing changes that add code to AgentSideEffects or AgentService.

microsoft/vscode · 56 tokens