Agent skill

Ultracode

by KonghaYao in KonghaYao/peri

Multi-agent workflow orchestration. An agent skill from KonghaYao/peri.

Apache-2.0Auto-check passedAgent Workflows

Install Ultracode

skills CLI
$ npx skills add KonghaYao/peri --skill ultracode -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install KonghaYao/peri ultracode --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/KonghaYao/peri.git skills-src && mkdir -p .claude/skills && cp -r skills-src/peri-middlewares/src/skills/builtin/skills/ultracode .claude/skills/ultracode && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
ultracode
GitHub stars
224
Token cost
~2k tokens
SKILL.md length
669 words
Files
1
Skills in repo
19
Repo updated
First seen
Licence
Apache-2.0

At a glance

Multi-agent workflow orchestration. An agent skill from KonghaYao/peri.

  • Works in 10 steps: Discover the Workflow tool → Write a workflow script → Example: Parallel Code Review → …
  • The user says ultracode
  • SKILL.md covers When to Use, How to Use, Prerequisites and Best Practices, plus 1 more section
  • Calls npx

What it does

Ultracode is an agent skill from KonghaYao/peri. Multi-agent workflow orchestration. Use when the user says "ultracode", "workflow", "parallel agents", "pipeline", or asks to orchestrate multiple agents for complex tasks. Also use when the user wants to break work into phases or parallel streams.

Its SKILL.md is about 2k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Agent Workflows, covering Subagents and Multi-agent orchestration. The repository describes itself as: Lightweight Rust Agent only use 50MB RAM, but Claude Code Plugin compatible, Dynamic Workflow, Goal, Artifacts, Free Web Search, full feature and better support! The licence is Apache-2.0.

When your agent uses it

  • The user says ultracode
  • Parallel agents
  • Asks to orchestrate multiple agents for complex tasks
  • The user wants to break work into phases

Example prompts

  • “ultracode”
  • “workflow”
  • “parallel agents”
  • “/ultracode”

Requirements

  • Node.js

Workflow steps

10 steps, taken from the step headings in SKILL.md.

  1. Discover the Workflow tool
  2. Write a workflow script
  3. Example: Parallel Code Review
  4. Example: Pipeline Data Processing
  5. Example: Phase Marking
  6. Example: Sub-Workflow Invocation
  7. Validate a saved script before running it
  8. Inspect saved workflow results with the CLI
  9. The workflow runs asynchronously
  10. Monitor progress

What it can do on your machine

Read from SKILL.md and the folder at commit d7ee444. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • npx

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    Links to these hosts (documentation or services it may open):

    • nodejs.org

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Ultracode loads about 2k tokens when it runs. Until then it costs about 65 tokens; SKILL.md has 669 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~65
When it runs · the whole SKILL.md, loaded when a task matches
~2k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from KonghaYao/peri at commit d7ee444, republished under its Apache-2.0 licence (© KonghaYao). 669 words, ~1,959 tokens.

Download SKILL.mdSave it as .claude/skills/ultracode/SKILL.md (or your agent's skills folder).
name
ultracode
description
Multi-agent workflow orchestration. Use when the user says "ultracode", "workflow", "parallel agents", "pipeline", or asks to orchestrate multiple agents for complex tasks. Also use when the user wants to break work into phases or parallel streams.
userInvocable
true
argumentHint
[task description or /ultracode]

Ultracode: Multi-Agent Workflow Orchestration

You have access to the Workflow tool (via SearchExtraTools → ExecuteExtraTool), which lets you orchestrate multiple agents working in parallel or pipeline phases.

When to Use

Use the Workflow tool when:

  • The task can be decomposed into independent parallel subtasks
  • The task benefits from a pipeline (phased execution where later phases depend on earlier results)
  • You need to explore multiple approaches simultaneously
  • The user explicitly asks for "ultracode", "workflow", or parallel execution

An explicit user request for this skill or a workflow takes precedence over the convenience guidelines below. Use the requested workflow even when direct execution would be faster; keep its scope small. If the required tool is unavailable, explain the blocker and obtain the user's choice before replacing the requested execution method.

When the user has not specified a workflow, prefer direct execution for:

  • Simple single-agent tasks (just do the work directly)
  • Tasks requiring tight sequential conversation (use normal tool calls)
  • Tasks that are faster to do inline than to script

How to Use

1. Discover the Workflow tool

The Workflow tool is a deferred tool. First search for it:

SearchExtraTools("workflow")

Then execute it:

ExecuteExtraTool("Workflow", {
  "script": "...your workflow script...",
  "args": {},
  "maxConcurrency": 3
})
2. Write a workflow script

Workflow scripts are JavaScript ESM modules using these primitives:

  • agent(prompt, options?) — Run a single agent. Returns the agent's output.
  • parallel([...factories]) — Run multiple agents concurrently. 入参为返回 promise 的零参工厂函数(thunks),不是 promise。 Returns array of results.
  • pipeline(items, ...stages) — 数据流水线:对 items 每个元素顺序执行所有 stage,stage=(prev, item, index) => result,stage 的返回值为下一个 stage 的 prev。返回与 items 等长的数组,出错位置为 null。
  • phase(title) — 切换当前阶段标记(发 phase_started/done 事件),后续 agent 自动带该阶段名。不接 fn,返回 undefined。
  • log(message) — Emit a log message visible in the workflow panel.
  • workflow(nameOrScriptPath, args?) — 运行子 workflow(一层嵌套上限)。workflow('sub-name', args) 按名字查找,workflow({ scriptPath: '...' }, args) 按文件路径加载。
3. Example: Parallel Code Review
javascript
export const meta = {
  name: 'parallel-review',
  description: 'Review code from multiple perspectives in parallel'
}

const [security, perf, bugs] = await parallel([
  () => agent('Review this code for security vulnerabilities. Focus on injection, auth, and data exposure.', { label: 'security', allowedTools: ['Read', 'Grep'] }),
  () => agent('Review this code for performance issues. Focus on N+1 queries, unnecessary allocations, and hot paths.', { label: 'performance', allowedTools: ['Read', 'Grep'] }),
  () => agent('Review this code for bugs. Focus on edge cases, error handling, and logic errors.', { label: 'bugs', allowedTools: ['Read', 'Grep'] }),
])

return { security, perf, bugs }

⚠️ parallel 入参必须是工厂函数 () => agent(...),不能直接写 agent(...)。 直接传 agent(...) 调用(已是 Promise、不可调用)会被 runtime 静默吞掉:parallel 把每个元素当函数调用,抛 TypeError 后被内部 catch 返回 null,workflow 以「假成功 completed + 全 null 返回值」结束、且不写 journal.jsonl。务必写成 () => agent(...)。

4. Example: Pipeline Data Processing
javascript
// pipeline(items, ...stages): 对每个数据项顺序跑 stage
const files = ['src/a.rs', 'src/b.rs']

const results = await pipeline(files,
  // stage1: 审查每个文件
  (file) => agent(`审查文件 ${file} 的代码质量`, { label: `review:${file}`, model: 'haiku' }),
  // stage2 (可选): 对 stage1 输出做二次处理,prev=上阶段结果
  (prev) => agent(`总结审查结果: ${prev}`, { label: 'summarize', model: 'haiku' }),
)

return results  // [{stage2_result_a}, {stage2_result_b}]
5. Example: Phase Marking
javascript
// phase(title): 切换阶段标记,后续 agent 自动带该阶段名
phase('Review')
const r1 = await agent('审查代码', { label: 'reviewer' })

phase('Fix')
const r2 = await agent('修复问题', { label: 'fixer' })

return { r1, r2 }
6. Example: Sub-Workflow Invocation
javascript
// workflow(nameOrScriptPath, args?): 运行子 workflow(一层嵌套上限)
const subResult = await workflow(
  { scriptPath: 'scripts/sub-check.mjs' },
  { threshold: 0.8 },
)
return subResult  // 子 workflow 的 return 值
7. Validate a saved script before running it

@peri-code/workflow provides a CLI preflight check for an existing workflow file:

bash
# 必须带显式版本号:npx 在全局已有同名 bin 时会静默复用旧版(0.1.x 无 CLI 子命令,
# 所有命令无输出且 exit 0,无任何报错),`@<version>` 才能强制使用 registry 目标版本。
npx -y @peri-code/workflow@0.2.0 validate scripts/parallel-review.mjs
# Machine-readable output for tooling:
npx -y @peri-code/workflow@0.2.0 validate scripts/parallel-review.mjs --json

Use it when a workflow script is already saved and non-trivial. It checks engine syntax and exports, rejects old workflow.agent(...)-style calls, requires export const meta = { name, description }, and warns when the script has no top-level return.

Do not create a file solely to validate an inline script passed directly to the Workflow tool.

Show full SKILL.md (288 more words)Show less
8. Inspect saved workflow results with the CLI

After a run, the runner persists state.json, agent journal entries, and extracted long outputs under .claude/workflow-runs/<run_id>/. The Read tool remains suitable for a known result file; use the built-in CLI when you need a complete report or need to discover runs:

bash
npx -y @peri-code/workflow@0.2.0 list
npx -y @peri-code/workflow@0.2.0 list --json
npx -y @peri-code/workflow@0.2.0 read <run_id>
npx -y @peri-code/workflow@0.2.0 read <run_id> --short --json

read and list search upward from the current directory for .claude/workflow-runs/. read restores extracted long outputs in its report.

9. The workflow runs asynchronously

The Workflow tool returns immediately with a run_id. You'll receive a notification when it completes. The results are saved to .claude/workflow-runs/<run_id>/state.json.

Use the Read tool or the read CLI command to examine the results after completion.

10. Monitor progress

Use /workflows to open the workflow panel and see real-time progress (phases, agents, token counts).

Prerequisites

The Peri host prefers its locally installed @peri-code/workflow Node bundle and automatically falls back to npx -y @peri-code/workflow@<version> (explicit version pin) when that bundle is unavailable. No global @peri-code/workflow installation is required. The Workflow tool requires Node.js (provides npx); if npx is unavailable, it returns an error:

npx is not available. Install Node.js (https://nodejs.org/) to enable workflow support.

In that case, tell the user to install Node.js and retry.

Best Practices

  • Keep agent prompts focused and specific
  • Use allowedTools to restrict each agent's capabilities
  • Use label to identify agents in the progress panel
  • Use phase() to organize work into named stages
  • Set maxConcurrency appropriately (default 3)
  • Cache results with journal/resume for expensive operations

Script Constraints

workflow 脚本运行在确定性沙箱中(保证 journal resume 一致性),以下运行时能力受限:

  • 禁止 Date.now() / new Date():时间戳不得在脚本内自行获取,需经 args 由调用方注入。使用会触发 failed: "Date.now()/new Date() is not available in workflow scripts"。
  • meta 必须同时含 name 和 description:export const meta 缺少任一字段时 workflow 立即 failed: "meta must include string name and description"。
  • 其他受限制的全局对象以 engine 运行时为准(如 Math.random 等非确定性 API 可能也受限,待补充完整清单)。

© KonghaYao, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in peri-middlewares/src/skills/builtin/skills/ultracode of KonghaYao/peri.

Open the folder on GitHubat commit d7ee444

Compare with similar skills

Ultracode next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Ultracode compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Ultracode this skillKonghaYao/peri224—~2kAutomated safety check: PassApache-2.0
Paseo Advisor Second Opiniongetpaseo/paseo20k1 repos~756Automated safety check: PassCustom licence
O2 Review Loopopenobserve/openobserve22k—~3.7kAutomated safety check: PassAGPL-3.0
Harness Agent Team Designerrevfactory/harness9.1k—~4.5kAutomated safety check: PassApache-2.0
ClawTeam Multi-Agent Swarmwin4r/ClawTeam-OpenClaw1.5k1 repos~2.9kAutomated safety check: PassMIT
Sub-Agent Delegationcodewhale-hq/Codewhale41k—~790Automated safety check: PassMIT

Similar skills

  • Launches one separate agent through Paseo to give a second opinion on the current task, with a self-contained briefing and no permission to edit files.

    20k GitHub starsUsed in 1 repo~756 tokens
    Agent WorkflowsAuto-check passed
  • O2 Review Loop

    openobserve/openobserve

    Splits a change into planner, coder and independent reviewer roles: you confirm a spec, a subagent implements it, and a separate reviewer checks each round's local WIP commit.

    22k GitHub stars~3.7k tokensUpdated today
    Agent WorkflowsAuto-check passed
  • Harness Agent Team Designer

    revfactory/harness

    Designs a project-specific agent harness: defines specialist agents, writes the skills they follow, picks an execution mode and model for each, and keeps the setup maintained.

    9.1k GitHub stars~4.5k tokensUpdated 11 days ago
    Agent WorkflowsAuto-check passed
  • ClawTeam Multi-Agent Swarm

    win4r/ClawTeam-OpenClaw

    Launches a swarm of specialist Hermes agents in git-worktree-isolated tmux windows with a kanban board and file-based inboxes, using built-in templates like hedge-fund and code-review.

    1.5k GitHub starsUsed in 1 repo~2.9k tokens
    Agent WorkflowsAuto-check passed
  • Sub-Agent Delegation

    codewhale-hq/Codewhale

    Guides when and how to split multi-step coding, research or verification work into focused sub-agent runs while the parent keeps integration and final checks.

    41k GitHub stars~790 tokensUpdated today
    Agent WorkflowsAuto-check passed
  • Parallel Agents

    xenitV1/Antigravity-Workflows

    Native multi-agent orchestration using Claude Code's Agent Tool.

    130 GitHub starsUsed in 9 repos~1.3k tokens
    Agent WorkflowsAuto-check passed

More from KonghaYao/peri

All 19 skills in this repo
  • Queries Langfuse traces, prompts, datasets and sessions, and analyzes local LLM gateway logs for requests, context growth, token use and cache hits.

    224 GitHub stars~4.3k tokensUpdated today
    Auto-check: notes
  • Audits recent agent conversation history and turns repeated failures and successes into testable harness improvement proposals that later audits can check.

    224 GitHub stars~3.5k tokensUpdated today
    Auto-check passed
  • Runs commands, reads and edits files, and copies data on remote machines through a single-file Node script that wraps the system ssh and scp, in Chinese.

    224 GitHub stars~924 tokensUpdated today
    Auto-check: warnings
  • Advisor Consultation

    KonghaYao/peri

    Sends a compact, redacted decision packet to a tool-free Opus advisor subagent when a task has high-risk trade-offs or stalled investigations, then weighs the answer.

    224 GitHub stars~1.3k tokensUpdated today
    Auto-check passed
  • Scheduled Tasks Cron

    KonghaYao/peri

    Registers, lists and removes recurring agent tasks with five-field cron expressions, and sets safety rules so a schedule is created only when the user clearly asks.

    224 GitHub stars~683 tokensUpdated today
    Auto-check passed
  • Verifies and repairs a feature by using the real Peri terminal UI as a user would, looping verify, decide, fix and review until a fresh round shows no blockers.

    224 GitHub stars~2.1k tokensUpdated today
    Auto-check: notes

Categories

Questions about Ultracode

What does Ultracode do?

Multi-agent workflow orchestration. An agent skill from KonghaYao/peri. Ultracode is an agent skill from KonghaYao/peri. Multi-agent workflow orchestration.

When should I use Ultracode?

Ultracode fits situations like: the user says ultracode; parallel agents; asks to orchestrate multiple agents for complex tasks; the user wants to break work into phases.

How do I install Ultracode in Claude Code?

Run `npx skills add KonghaYao/peri --skill ultracode -a claude-code`. Or copy the skill folder (peri-middlewares/src/skills/builtin/skills/ultracode in KonghaYao/peri) into .claude/skills/ultracode in your project. Claude Code loads it when a task matches its description.

How do I install Ultracode in Codex?

Run `npx skills add KonghaYao/peri --skill ultracode -a codex`. Or copy the skill folder (peri-middlewares/src/skills/builtin/skills/ultracode in KonghaYao/peri) into .agents/skills/ultracode in your project. Codex loads it when a task matches its description.

Can I use Ultracode in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add KonghaYao/peri --skill ultracode -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/ultracode, .gemini/skills/ultracode, .github/skills/ultracode and .opencode/skills/ultracode in your project.

What does Ultracode need to run?

Going by SKILL.md and its folder, Ultracode needs the command-line tools its instructions call (npx). Our summary lists: Node.js.

Does Ultracode access the network?

SKILL.md names 1 domain. As links in the text: nodejs.org. This is read from the text; nothing was executed.

Is Ultracode safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Ultracode use?

Ultracode is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Ultracode use?

About 2k tokens (SKILL.md is roughly 7.8k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Ultracode?

Skills that share tags, products or a category with Ultracode: Paseo Advisor Second Opinion (getpaseo/paseo, 20k stars), O2 Review Loop (openobserve/openobserve, 22k stars), Harness Agent Team Designer (revfactory/harness, 9.1k stars) and ClawTeam Multi-Agent Swarm (win4r/ClawTeam-OpenClaw, 1.5k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Ultracode?

KonghaYao (a GitHub user) maintains it in KonghaYao/peri, which has 224 GitHub stars. The repository holds 19 skills in this directory. The repository was last updated on October 9, 2026.

Source: KonghaYao/peri on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.