← 返回 Skills 市场
1844
总下载
1
收藏
4
当前安装
1
版本数
在 OpenClaw 中安装
/install optimize-context
功能描述
Automatically monitors and optimizes conversation context to prevent prompt size errors by extracting key points and clearing excess history.
安全使用建议
This package implements what it says (summarizing and splitting large tasks) but you should review and consider the following before installing:
- Data persistence: The skill extracts 'facts' from conversations and writes them to files (MEMORY.md and context-summary-*.md) in the workspace. If conversations contain sensitive information, that data will be stored on disk in plain text. Decide whether you are comfortable with this and consider encrypting or relocating the memory files.
- Path inconsistency: The SKILL.md installation path (~/.openclaw/workspace) and the code's hardcoded workspace path (/home/blackworm/.openclaw/workspace) differ — verify and correct paths before installing so files are created where you expect.
- Automatic behavior: The system includes background monitoring and scheduled optimization. If you do not want automated runs, disable scheduling in task_processing_config.json or avoid running the monitor/cron integration.
- Test in a sandbox: Install and run the skill in a disposable or non-production workspace first to observe what files it creates and confirm behavior.
- Minimal privileges: The skill does not request external credentials or network access, which is good. Still inspect MEMORY.md and summary files after a test run and remove any sensitive data you do not want persisted.
If you want, I can point out the exact lines that write files, show how to change the hardcoded path to a dynamic location, or produce a checklist to safely sandbox and audit this skill before use.
功能分析
Type: OpenClaw Skill
Name: optimize-context
Version: 1.0.0
The OpenClaw AgentSkills skill bundle is designed for context optimization and large task processing. All file system operations are confined to the OpenClaw workspace (`/home/blackworm/.openclaw/workspace`) and its designated memory/configuration files (`task_processing_config.json`, `MEMORY.md`, `memory/context-summary-*.md`). There is no evidence of data exfiltration, malicious execution (e.g., `eval`/`exec` of untrusted input, remote payload execution), persistence mechanisms, or prompt injection attempts in the markdown files. The skill's behavior is clearly aligned with its stated purpose of managing conversation context and splitting tasks.
能力评估
Purpose & Capability
Name/description (context optimization / task processing) aligns with the included code: modules implement summarization, fact extraction, context cleanup, and task splitting. There are no requested credentials or unrelated binaries.
Instruction Scope
The runtime instructions and code read conversation messages, extract facts, and write persistent summary/memory files (MEMORY.md and context-summary-*.md). The SKILL.md instructs extraction into ~/.openclaw/workspace, but most code uses a hardcoded path (/home/blackworm/.openclaw/workspace) — a clear inconsistency. The skill's behavior includes automatic background monitoring and scheduled optimization; these actions will collect and persist conversation content to disk, which is outside a purely in-memory optimization expectation and increases privacy risk.
Install Mechanism
No formal install spec provided (instruction-only + bundled code). SKILL.md suggests running npm install in the skill directory, but the skill's package.json shows no external dependencies; running npm install is low risk but may be unnecessary. No third-party downloads or remote URLs are used.
Credentials
The skill requests no environment variables or external credentials, which is appropriate. However, it persists extracted 'facts' and summaries to disk (workspace/memory files) without encryption or explicit user consent. Because the skill automatically extracts and stores potentially sensitive user-provided content, that persistent storage is a material privilege not declared in metadata and should be considered when evaluating proportionality.
Persistence & Privilege
The skill is not configured always:true and won't be force-enabled, but it does create and modify files in the OpenClaw workspace (memory directory, MEMORY.md, summary files) and includes code for background monitoring and scheduled runs. It does not modify other skills' configurations or request elevated system privileges, but its ability to persist conversation data increases its blast radius if installed.
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 optimize-context - 安装完成后，直接呼叫该 Skill 的名称或使用
/optimize-context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 Context Optimizer & Task Processing Skills Package.
- Adds automated context optimization to prevent "prompt too large" errors.
- Introduces a Task Processor to split and handle large tasks automatically.
- Includes background monitoring for conversation length with configurable thresholds.
- Provides `/optimize-context` and `/process-task` commands for manual control and automation.
元数据
常见问题
Optimize Context 是什么？
Automatically monitors and optimizes conversation context to prevent prompt size errors by extracting key points and clearing excess history. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 1844 次。
如何安装 Optimize Context？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install optimize-context」即可一键安装，无需额外配置。
Optimize Context 是免费的吗？
是的，Optimize Context 完全免费（开源免费），可自由下载、安装和使用。
Optimize Context 支持哪些平台？
Optimize Context 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 Optimize Context？
由 blackworm（@blackworm）开发并维护，当前版本 v1.0.0。
推荐 Skills