← 返回 Skills 市场

Hermes Agent Skill

Name: Hermes Agent Skill
Author: wihy

作者 will · GitHub ↗ · v2.1.1 · MIT-0

cross-platform ✓ 安全检测通过

2078

总下载

当前安装

版本数

在 OpenClaw 中安装

/install hermes-agent-v2

功能描述

NousResearch Hermes Agent CLI integration. Core capabilities: - Self-improving skill system - Persistent memory (FTS5 + LLM summaries) - Sub-agent delegation...

使用说明 (SKILL.md)

Hermes Agent Skill

Call NousResearch Hermes Agent via CLI.

Prerequisites

Hermes CLI must be installed. See https://github.com/NousResearch/hermes-agent for installation.

Usage

Quick Q&A

hermes run "your question" --non-interactive --no-stream

With Context

hermes run "question" --context-file ./ctx.md --non-interactive

Sub-agent Delegation

hermes run "use delegate_task to: research topic" --non-interactive --no-stream

Memory

hermes memory search "keyword"
hermes memory notes list
hermes memory notes add "note content"

Skills

hermes skills list
hermes skills create "name" --description "desc"

Status

hermes status
hermes doctor

Configuration

Edit ~/.hermes/.env to set your LLM provider key. Edit ~/.hermes/config.yaml to set model and provider.

File Access Scope

This skill only invokes hermes CLI commands. No direct file operations. Hermes Agent operates within ~/.hermes/ only.

安全使用建议

This skill itself only runs the Hermes CLI — it does not request secrets or write files outside ~/.hermes/ — but Hermes can perform browser automation, spawn sub-agents, run code, and will use any LLM provider key you put in ~/.hermes/.env. Before installing/use: (1) verify you trust and install Hermes from the official repository URL referenced in SKILL.md; (2) review the contents of ~/.hermes/ and the .env file so you know which API keys are stored and used; (3) avoid putting high-privilege credentials in ~/.hermes/.env (use scoped or test keys); (4) consider running Hermes and this skill in a sandbox or with manual approval for sensitive operations (delegation, code execution, browser automation); (5) when in doubt, test with a non-sensitive account and inspect Hermes' behavior first.

功能分析

Type: OpenClaw Skill Name: hermes-agent-v2 Version: 2.1.1 The skill bundle is a standard CLI integration for the NousResearch Hermes Agent. It provides documentation and command examples for interacting with the 'hermes' CLI tool, such as 'hermes run' and 'hermes memory'. While the underlying tool possesses high-risk capabilities like code execution and browser automation, these are consistent with its stated purpose as an AI agent, and the SKILL.md file contains no evidence of malicious intent, data exfiltration, or prompt injection.

能力评估

✓ Purpose & Capability

Name/description claim a Hermes CLI integration and the SKILL.md + _meta.json consistently instruct calling the `hermes` CLI; required binaries/configs align with that purpose (no unrelated credentials or tools requested).

ℹ Instruction Scope

SKILL.md only instructs invoking Hermes CLI commands (run, memory, skills, status). It does not itself instruct reading arbitrary host files or requesting extra env vars. However, those Hermes commands can accept context files and cause sub-agents, browser automation, or code execution — so the runtime behavior depends on the external Hermes binary and the data you pass it.

✓ Install Mechanism

Instruction-only skill with no install spec and no included code — lowest-risk packaging. The skill relies on an already-installed `hermes` CLI (meta.json suggests a `which hermes` check).

ℹ Credentials

The skill declares no required env vars or credentials. It does advise editing `~/.hermes/.env` to set an LLM provider key — reasonable for a CLI agent, but that means credentials live in the user's home config (outside the skill). The skill itself does not request unrelated secrets.

✓ Persistence & Privilege

always: false and no special persistence requested. Model invocation is allowed (platform default). Be aware: allowing autonomous invocation plus Hermes' delegation/code-execution features increases the potential impact of actions Hermes is asked to perform, but that is a characteristic of the external tool rather than the skill bundle.

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 hermes-agent-v2
安装完成后，直接呼叫该 Skill 的名称或使用 /hermes-agent-v2 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

v2.1.1

- Major simplification: Documentation and scripts have been removed; only core hermes CLI integration remains. - All helper scripts and reference guides deleted; only essential usage and installation instructions are included. - Security and file access descriptions clarified: skill only calls hermes CLI and accesses ~/.hermes/. - LICENSE file added; documentation now focuses on basic usage and configuration.

v2.1.0

1、SKILL.md 添加 security 字段，明确文件访问范围 2、error_callback 改为仅输出到 stderr

v2.0.0

**Hermes Agent v2.0.0 – Major upgrade to portable, scriptable integration** - Rebuilt skill as a fully portable, script-driven Hermes Agent integration: no hardcoded paths, easy migration between Claw/WorkBuddy instances - Added detailed Chinese-language setup/configuration guide with automated install (`install_hermes.sh`), API key setup, and troubleshooting - Introduced wrapper and delegate CLI scripts for unified command handling and subagent delegation - Provided comprehensive CLI reference, best practices, and cost guidance - File structure and documentation now focused on CLI/multi-platform workflows; legacy workspace seeding/memory/promotion files removed

元数据

Slug hermes-agent-v2

版本 2.1.1

许可证 MIT-0

累计安装 1

当前安装数 1

历史版本数 3

常见问题