agentleFS
timursafiullintimursafiullin/aod/skills/wave/SKILL.md

wave

Skill0 starsChanged 29 days agoApache-2.0View on GitHub

Delegate several independent ready AOD tasks to Codex in parallel, one per worktree slot, after checking dependencies and that their scopes do not overlap. Use when two or more tasks are ready and the user wants throughput.

4 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

5 rules. 0 are common, 1 is distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    every dependency is done;

    line 14
  2. Very specific:

    Run it as a background shell command, so you are notified when every job has ended (do not poll): aod wave run [--max N].

    line 19
  3. Very specific:

    Read the result's ran list.

    line 22
  4. Distinctive:

    Verify and close tasks one at a time.

    line 27
  5. Very specific:

    Do not run /aod:delegate for a task while a wave is running.

    line 32

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Parallel wave

A wave runs independent tasks at the same time, one per slot. CLI: aod means node "${CLAUDE_PLUGIN_ROOT}/scripts/aod.mjs".

  1. Each task in the wave must be ready after /aod:preflight. The packet depends_on lists real dependencies, and the scope globs are as narrow as the work allows. Overlapping scopes never run together.
  2. Plan the wave: aod wave plan [--max N]. It picks ready Codex tasks in id order:

    • every dependency is done;
    • the scope overlaps neither another task in the wave nor any task still holding a slot;
    • there is at most one task per free slot.

    deferred explains why each other task waits. Show the user the wave in one line before running it.

  3. Run it as a background shell command, so you are notified when every job has ended (do not poll): aod wave run [--max N]. Slots are created first, then the delegations start together. Each one follows the routing, exactly like /aod:delegate.
  4. Read the result's ran list. Each entry has the same meaning as a /aod:delegate result:

    • review -> /aod:verify it;
    • blocked -> answer the question;
    • failed -> read stderr.log.

    Verify and close tasks one at a time. Each merge resets that slot and moves the base branch.

  5. Exit 5 means Codex is unavailable or out of quota. Every affected task is back to ready. Stop and ask the user (/aod:decision).
  6. Once the wave's tasks are closed, aod wave plan again. Tasks that depended on them are now eligible.

Do not run /aod:delegate for a task while a wave is running. A slot it picks may be one the wave assigned. Parallelism multiplies Codex usage: with the max preset, mention it before running waves of 3 or more.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.