ci-mockup-figure
CommandCreate space-efficient paper and proposal figures (HTML mockups, TikZ, or skia-canvas) from tool selection through LaTeX insertion
@yzhao062Install CommandDevelopment
Command
/prun
Run prun: parallel delegation fan-out on Agy workers (the session coordinates)
$ npx -y skills add yzhao062/anywhere-agents --agent claude-code
Command definition
prun.mddescription: "Run prun: parallel delegation fan-out on Agy workers, or Codex or Grok when you name it (the session coordinates)"
argument-hint: "[task description or context]"
Read and follow the skill definition. Look for it at `skills/prun/SKILL.md` first, then `.claude/skills/prun/SKILL.md`, then `.agent-config/repo/skills/prun/SKILL.md`.
Command arguments from the slash invocation: `$ARGUMENTS`
Treat the command arguments as the task to fan out. prun decomposes the task into independent units and runs many of them in parallel on Agy workers (Gemini through the Antigravity CLI, on the separate Google AI pool), never on this session. Do not spawn Sonnet or any other Claude subagent, or a Workflow, for a unit: those bill the same Claude account this session runs on. Every unit runs on Agy by default. A unit runs on Codex only when the user's request names Codex for that unit or a batch; this session never picks Codex on its own, not even as a fallback when Agy is short. A Codex unit goes through `scripts/dispatch-task.sh` (`dispatch-task.ps1` on Windows), and its model and effort come from the registry's `prun-worker` row for Codex. A unit runs on Grok only when the user's request names Grok for that unit or a batch; this session never picks Grok on its own, not even as a fallback when Agy is short (when Agy fails with `preflight timed out after 60s` or `Antigravity preflight failed`, exit 124 or 70, offer Grok to the user). A Grok unit goes through `scripts/dispatch-task-grok.py`, and its model and effort come from the registry's `prun-worker` row for Grok (see `skills/prun/references/grok-executor.md`). Each unit runs unattended in a scratch dir or throwaway clone and gets follow-up turns while slower units finish. Units may read or write code; code-writing units run in a throwaway local clone, workers never commit or push, and this session plus the user are the final integration gate. This session gathers the results, reviews each diff, and integrates. Choose the worker count from the actual dependency graph; do not impose an arbitrary two- or three-worker cap.
Read more
description: "Run prun: parallel delegation fan-out on Agy workers, or Codex or Grok when you name it (the session coordinates)" argument-hint: "[task description or context]"
Read and follow the skill definition. Look for it at `skills/prun/SKILL.md` first, then `.claude/skills/prun/SKILL.md`, then `.agent-config/repo/skills/prun/SKILL.md`.
Command arguments from the slash invocation: `$ARGUMENTS`
Treat the command arguments as the task to fan out. prun decomposes the task into independent units and runs many of them in parallel on Agy workers (Gemini through the Antigravity CLI, on the separate Google AI pool), never on this session. Do not spawn Sonnet or any other Claude subagent, or a Workflow, for a unit: those bill the same Claude account this session runs on. Every unit runs on Agy by default. A unit runs on Codex only when the user's request names Codex for that unit or a batch; this session never picks Codex on its own, not even as a fallback when Agy is short. A Codex unit goes through `scripts/dispatch-task.sh` (`dispatch-task.ps1` on Windows), and its model and effort come from the registry's `prun-worker` row for Codex. A unit runs on Grok only when the user's request names Grok for that unit or a batch; this session never picks Grok on its own, not even as a fallback when Agy is short (when Agy fails with `preflight timed out after 60s` or `Antigravity preflight failed`, exit 124 or 70, offer Grok to the user). A Grok unit goes through `scripts/dispatch-task-grok.py`, and its model and effort come from the registry's `prun-worker` row for Grok (see `skills/prun/references/grok-executor.md`). Each unit runs unattended in a scratch dir or throwaway clone and gets follow-up turns while slower units finish. Units may read or write code; code-writing units run in a throwaway local clone, workers never commit or push, and this session plus the user are the final integration gate. This session gathers the results, reviews each diff, and integrates. Choose the worker count from the actual dependency graph; do not impose an arbitrary two- or three-worker cap.
Ships withanywhere-agents
One config for every agent — Claude Code, Codex, and whatever comes next. Start with effective defaults. Add packs, small bundles of rules, skills, or permissions, as you need them. One AGENTS.md drives every agent in every repo on every machine.
2616 skills1 agent
Stars on GitHub
Momentum
- stars in total
- 261
Other commands on anywhere-agents.
@yzhao062Install Commandeditable-figure
CommandDesign concise overview, mechanism, or workflow figures as editable PowerPoint objects
@yzhao062Install Commandmy-router
CommandDetect the work type (papers, proposals, code, figures, admin) and dispatch to the right domain skill
@yzhao062Install Commandreadme-polish
CommandAudit a GitHub README and rewrite it with modern patterns for a scannable ten-second skim
@yzhao062Install CommandMore commands from other plugins.
Auto-invoked
Auto-invoked
build
CommandImplement tasks incrementally — build, test, verify, commit. Add "auto" to run the whole plan in one approved pass.
@addyosmaniInstall Command