← 返回 Skills 市场
Agent Oversight
作者
1477009639zw-blip
· GitHub ↗
· v1.0.0
· MIT-0
118
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install agent-oversight
功能描述
Comprehensive AI agent oversight and management skill. Monitors sub-agents, manages file edit coordination, logs failures, kills hung sessions, and maintains...
使用说明 (SKILL.md)
Agent Oversight
Manages and monitors AI sub-agents. Keeps Beta's workspace clean, coordinated, and error-free.
Usage
python3 oversight.py --status
python3 oversight.py --list-sessions
python3 oversight.py --kill-hung
Features
- Session monitoring (active/idle/hung)
- File edit coordination (prevents conflicts)
- Failure logging to memory/learnings.md
- Automatic kill for >5min hung sessions
- Coordination rules enforcement
Coordination Rules
- Read file before editing
- Exact oldText matching required
- One agent per file at a time
- Log all failures to learnings.md
安全使用建议
Do not install or enable this skill expecting the advertised oversight features: the shipped code only reports status and lists sessions via an external 'openclaw' CLI. Ask the author for a complete implementation (show the kill/hung-session logic, file-locking coordination code, and explicit logging behavior), and for details about the 'openclaw' command (what it requires, what data it returns, and whether it needs credentials). Until those are provided, treat the skill as incomplete — test it in a safe environment and verify the external CLI is trusted before granting it access to agent session management.
功能分析
Type: OpenClaw Skill
Name: agent-oversight
Version: 1.0.0
The agent-oversight skill is a utility for managing OpenClaw sub-agent sessions. The oversight.py script uses subprocess to safely call the local 'openclaw' CLI for session listing and provides status updates. While the SKILL.md documentation mentions features like killing hung sessions that are not yet fully implemented in the provided code, there are no indicators of malicious intent, data exfiltration, or prompt injection.
能力评估
Purpose & Capability
The name/description claim session killing, file-edit coordination, failure logging, and rule enforcement. The only required runtime binary is python3 and the included code merely prints status and runs 'openclaw sessions list --json'. The claimed capabilities (automatic kill, file locking, logging to learnings.md) are not implemented in the shipped code, an incoherence between purpose and actual capability.
Instruction Scope
SKILL.md instructs commands including --kill-hung and promises behavior like logging and file coordination, but oversight.py implements only --status and --list-sessions; --kill-hung and any file I/O/logging behavior are absent. The runtime instructions therefore overstate what the agent will do.
Install Mechanism
No install spec is provided (instruction-only with a single Python script). This is low-risk in terms of arbitrary downloads or on-disk installers.
Credentials
The skill declares no required environment variables or credentials, which is consistent with the tiny script. However, the script invokes an external 'openclaw' CLI without declaring what access or credentials that CLI needs — the reviewer should confirm what that CLI does and whether it requires privileged access or tokens.
Persistence & Privilege
always is false and the skill is user-invocable. It does not request persistent presence or modify other skills' configs in the provided 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 agent-oversight - 安装完成后，直接呼叫该 Skill 的名称或使用
/agent-oversight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 agent-oversight skill, providing robust monitoring and management for AI sub-agents.
- Monitors active, idle, and hung agent sessions.
- Coordinates file edits to prevent conflicts between agents.
- Automatically kills sessions hung for more than 5 minutes.
- Logs all failures to learnings.md for review and continuous improvement.
- Enforces coordination rules: read before editing, require exact text matches, one agent per file, and detailed failure tracking.
元数据
常见问题
Agent Oversight 是什么？
Comprehensive AI agent oversight and management skill. Monitors sub-agents, manages file edit coordination, logs failures, kills hung sessions, and maintains... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 118 次。
如何安装 Agent Oversight？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install agent-oversight」即可一键安装，无需额外配置。
Agent Oversight 是免费的吗？
是的，Agent Oversight 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Agent Oversight 支持哪些平台？
Agent Oversight 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 Agent Oversight？
由 1477009639zw-blip（@1477009639zw-blip）开发并维护，当前版本 v1.0.0。
推荐 Skills