← 返回 Skills 市场

Agency Agents Main

Name: Agency Agents Main
Author: keda118228-dev

作者 keda118228-dev · GitHub ↗ · v1.0.0 · MIT-0

cross-platform ⚠ suspicious

总下载

当前安装

版本数

在 OpenClaw 中安装

/install agency-agents-main

功能描述

The Agency - AI role switcher. Activate specialized professional roles (frontend developer, sales agent, marketing specialist, etc.) for different tasks. Use...

使用说明 (SKILL.md)

🎭 The Agency - Role Switcher

Access specialized AI professional roles from the agency-agents collection.

Available Roles

Roles are organized by domain in: ~/.agents/skills/agency-agents-main/

Engineering (engineering/):

frontend-developer, backend-architect, senior-developer, software-architect
data-engineer, security-engineer, devops-automator, sre
mobile-app-builder, embedded-firmware-engineer, solidity-smart-contract-engineer
technical-writer, code-reviewer, database-optimizer, rapid-prototyper
...and 15+ more

Business (sales/, marketing/, support/, finance/):

Sales agent, marketing specialist, support agent, financial analyst

Specialized (specialized/, strategy/, design/):

Various niche professional roles

Activation

When user asks to switch to a role, read the corresponding .md file and adopt that persona.

Format: ~/.agents/skills/agency-agents-main/{category}/{filename}.md

Categories:

engineering/ → engineering-{role}.md
sales/ → sales-{role}.md
marketing/ → marketing-{role}.md
finance/ → finance-{role}.md
support/ → support-{role}.md
product/ → product-{role}.md
design/ → design-{role}.md
strategy/ → strategy-{role}.md
specialized/ → specialized-{role}.md

Quick Reference

Category	Path Pattern	Example
Engineering	engineering/engineering-{role}.md	engineering/engineering-frontend-developer.md
Sales	sales/sales-{role}.md	sales/sales-account-executive.md
Marketing	marketing/marketing-{role}.md	marketing/marketing-content-strategist.md
Support	support/support-{role}.md	support/support-technical-support-specialist.md

Role Switching Protocol

User requests role switch (e.g., "switch to frontend developer")
Identify category and role name
Read the corresponding .md file
Adopt the role's personality, tone, and workflow
Confirm activation to the user

Note

This skill acts as a meta-skill. The actual role content is defined in the .md files within the agency-agents directory.

安全使用建议

This package appears to be a large collection of persona markdown files and a SKILL.md that tells the agent to read persona .md files and adopt roles — that by itself is coherent and low-risk. However, the bundle also contains several shell scripts (install.sh, convert.sh, integrations/mcp-memory/setup.sh, lint-agents.sh) even though no install step is declared. Before installing or executing anything: 1) Inspect the contents of the shell scripts (look for curl/wget, git clone, tar/extract, rm -rf, sudo, chmod, ssh/scp, or commands contacting external hosts). 2) If you need to run any of the scripts, run them in a sandboxed environment (VM or container) and review network activity. 3) Confirm the skill's source and repository origin — there's no homepage listed and the source is 'unknown'; prefer packages from a known repository or author. 4) Note that the SKILL.md references a path in your home (~/.agents/skills/agency-agents-main/) — ensure the agent only reads the intended directory and that permissions limit broader filesystem access. If you cannot review the scripts yourself, treat this skill as untrusted and avoid executing any bundled shell scripts.

功能分析

Type: OpenClaw Skill Name: agency-agents-main Version: 1.0.0 This skill bundle is a highly professional and comprehensive library of specialized AI agent personas and deployment utilities. The core functionality is defined in SKILL.md as a 'role switcher,' and the accompanying shell scripts (install.sh, convert.sh) are legitimate tools for deploying these personas into various AI environments like Claude Code, GitHub Copilot, and Cursor. The extensive collection of markdown files across multiple divisions (Engineering, Sales, Marketing, etc.) provides detailed system prompts that are strictly aligned with their stated professional purposes. No evidence of data exfiltration, malicious execution, or harmful prompt injection was found.

能力标签

cryptorequires-walletcan-make-purchasescan-sign-transactionsrequires-oauth-tokenrequires-sensitive-credentials

能力评估

ℹ Purpose & Capability

Name/description and SKILL.md match: the skill provides many persona .md files and instructs the agent to read them and adopt personas. Declared requirements are minimal (no env vars, no binaries), which is proportionate for a role-switcher. However the bundle contains multiple shell scripts (scripts/install.sh, scripts/convert.sh, integrations/mcp-memory/setup.sh, scripts/lint-agents.sh) and ~230 files — this is more than an instruction-only skill would typically need. The presence of install/utility scripts is a mismatch with the 'instruction-only' install metadata and should be examined.

ℹ Instruction Scope

The SKILL.md instructs the agent to read persona files from a specific path (~/.agents/skills/agency-agents-main/{category}/{file}.md) and adopt that persona. That scope is narrow and consistent with the purpose. But SKILL.md references a filesystem path within the user's home that the skill did not list under required config paths; the agent will need read access to that path. The instructions do not request external endpoints or additional credentials.

⚠ Install Mechanism

No install specification is declared, yet the package contains executable shell scripts (install.sh, convert.sh, setup.sh). Because there is no declared install step, it's unclear whether those scripts are intended to be executed by the agent or left for manual use by an operator. Shell scripts bundled with an otherwise 'instruction-only' skill increase risk: they can contain arbitrary commands (network downloads, filesystem changes, privilege changes). The SKILL.md does not instruct executing them, but their presence is an unexplained deviation that warrants manual review before running anything from the package.

✓ Credentials

The skill declares no environment variables, credentials, or special config paths. That aligns with its purpose as a local persona switcher. The only notable point is SKILL.md's explicit reference to a home-directory path (~/.agents/skills/agency-agents-main/) even though 'required config paths' was empty — reading files in that directory is necessary for operation, so the path should have been declared.

✓ Persistence & Privilege

The skill does not request permanent inclusion (always: false) and does not declare elevated privileges. Autonomous invocation (disable-model-invocation: false) is the platform default; nothing in the manifest escalates privilege. The risk center is the bundled scripts (see install_mechanism): if executed they could write/modify files, but the SKILL.md itself only instructs reading persona .md files.

如何使用

确保已安装 OpenClaw（本地或 Docker 部署）
在对话框中输入安装命令：/install agency-agents-main
安装完成后，直接呼叫该 Skill 的名称或使用 /agency-agents-main 触发
根据 Skill 的参数说明提供必要输入，即可获得结构化输出

版本历史

v1.0.0

- Initial release of The Agency - Role Switcher. - Activate specialized professional roles on request (e.g., frontend developer, sales agent, marketing specialist). - Supports a wide range of domains: engineering, business, specialized fields, and more. - Uses role descriptions from organized `.md` files for accurate persona switching. - Provides clear instructions and structure for activating and confirming new roles.

元数据

Slug agency-agents-main

版本 1.0.0

许可证 MIT-0

累计安装 1

当前安装数 1

历史版本数 1

常见问题