Token导航 LogoToken导航TokenDH.com
开发需要联网clawhub未标认证来源可访问clear审计通过

cpa-agents注册会计师 Agent 人

Agent Skill

cpa-agents 用于处理 GitHub 仓库、Issue、Pull Request 和代码协作信息,适合在 OpenClaw 中需要围绕仓库状态、代码变更或协作事项进行整理时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

3,036

周安装

124

GitHub Stars

公开资料未说明

下载量

972
OpenClaw

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

MIT-0

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:cpa-agents(注册会计师 Agent 人)
来源仓库:https://github.com/jose-compu/cpa-agents
安装命令:
openclaw skills install cpa-agents
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install cpa-agents

简介

使用并发进程代数模式(例如并行任务、分支修复循环、扇出比较和会话状态跟踪)管理 AI 代理工作流程。

SKILL.md

CPA Agents for OpenClaw

Version: 2.0.0

Use concurrent process algebra for practical multi-agent orchestration in OpenClaw. This skill is focused on production workflows: parallel execution, branch-fix loops, model fan-out, and session status introspection.

Install

1) Install library dependency

npm install cpa-agents

2) Create skill entrypoint

import { createOpenClawSkill } from "cpa-agents/adapters/openclaw";

export default createOpenClawSkill();

3) Runtime prerequisites

  • OpenClaw Gateway is running and reachable.
  • Skill host supports async command handlers.
  • Session context provides appendEvent for trace streaming.

Advanced operators

Use these when you need stronger control over failures, rollback, and data lineage.

Undo / rollback (invertible + saga)

  • invertible(forward, undo) defines a step plus compensating action.
  • saga([...steps]) executes steps in order; on failure, runs undo in reverse order.
  • Best for workflows with side effects (branch creation, file writes, merges, etc.).

Pattern:

import { invertible, saga } from "cpa-agents";

const workflow = saga([
  invertible(createBranch, deleteBranch),
  invertible(writeCode, revertCode),
  invertible(runValidation, cleanupValidation),
]);

Provenance (converse)

  • converse(R, inverseFn) is relational provenance, not operational undo.
  • Use it to answer: "which input(s) could have produced this output?"
  • Keep this separate from rollback logic (invertible/saga).

Pattern:

import { rel, converse } from "cpa-agents";

const generate = rel("generate", async (prompt: string) => [prompt + "-out"]);
const provenance = converse(generate, async (output: string) => [output.replace("-out", "")]);

Guarded flow (guard, guardValue, ifThenElse)

  • Use guard(...) to block unsafe execution.
  • Use guardValue(...) when a required value must exist.
  • Combine with ifThenElse(...) for conditional routing.

Reliability (retryWithBackoff, timeout, or)

  • retryWithBackoff for transient failures.
  • timeout(ms, process) to bound execution.
  • or(primary, fallback) for fallback routing when primary fails.

Practical guidance: undo vs converse

  • Use undo (invertible + saga) when you must reverse side effects.
  • Use converse for provenance/audit and lineage queries.
  • Typical production setup uses both:

- saga during execution - converse for post-run explainability

Commands

parallel

Run independent tasks concurrently.

{
  "tasks": [
    "analyze failing tests",
    "draft fix proposal",
    "write migration notes"
  ],
  "timeout": 300000
}

branch-fix

Run one task, and if errors are returned, branch into fix flow and continue.

{
  "task": "implement auth middleware and resolve lint issues",
  "timeout": 300000
}

fan-out

Run the same task across multiple models and merge outputs.

{
  "task": "propose API surface for agent orchestration",
  "models": ["model-a", "model-b"],
  "timeout": 300000
}

status

Return the current CPA process tree/status for the session.

{}

Recommended prompts

  • "Run parallel: audit security risks, propose patch plan, generate tests"
  • "Branch-fix this refactor until no type errors remain"
  • "Fan-out this architecture question across two models and compare"
  • "Show cpa status"
  • "Execute as saga: create branch, patch files, validate, rollback on failure"
  • "Show provenance: what input likely produced this output?"

Troubleshooting

  • Gateway not connected:

- Start OpenClaw Gateway and retry.

  • Unknown command:

- Use one of: parallel, branch-fix, fan-out, status.

  • Timeout:

- Increase timeout in command args.

  • Missing session events:

- Ensure session context is present and appendEvent is enabled.

  • Rollback did not occur:

- Verify steps are wrapped with invertible(...) and executed with saga(...).

  • Provenance unclear:

- Provide an explicit inverseFn when defining converse(...).

适合场景

01

OpenClaw 用户查找和安装 Skill 时

02

用户想查找某类 Agent Skill 时

03

需要根据任务场景推荐可安装能力包时

04

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

OpenClaw

97.29%
按下载量换算946

安全审计

VirusTotal

通过

ClawScan

通过

Static analysis

通过

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills