← 返回 Skills 市场
Reliable Tool Context
作者
Lourenço Maciel
· GitHub ↗
· v1.0.0
324
总下载
1
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install reliable-tool-context
功能描述
Build reliable tool context from command output using artifacts and compact reproducible code queries.
安全使用建议
This skill is internally consistent: it simply instructs the agent how to use the sift-gateway CLI to capture and query command output. Before installing, verify the sift-gateway package provenance (the uv registry entry or the GitHub project) and prefer official release sources (or the pipx alternative mentioned). Remember that running this skill gives the agent a binary (sift-gateway) that can execute arbitrary CLI commands you allow it to run — review your agent's permissions and policies so artifacts or command outputs you consider sensitive are not captured or leaked. If you need stronger guarantees, inspect the sift-gateway code or run it in a constrained environment (container or sandbox) before enabling the skill broadly.
功能分析
Type: OpenClaw Skill
Name: reliable-tool-context
Version: 1.0.0
The skill bundle facilitates the execution of arbitrary shell commands and Python code via the 'sift-gateway' utility (SKILL.md). While these capabilities are central to the stated purpose of capturing and querying command output artifacts, they represent a high-risk attack surface for Remote Code Execution (RCE). The instructions in SKILL.md guide the agent to use a Python 'code' flag to process data, which is functionally equivalent to an eval/exec pattern. No evidence of intentional malice, data exfiltration, or harmful prompt injection was found.
能力评估
Purpose & Capability
Name/description describe capturing CLI output and querying artifacts; the only required binary is sift-gateway and the SKILL.md instructs exactly how to use it. The declared install (uv package sift-gateway) matches the claimed capability.
Instruction Scope
Runtime instructions are narrowly focused on running sift-gateway run/code, handling pagination, resolving root paths, and protecting against pasting raw payloads. There are no directives to read unrelated files, environment variables, or to transmit data to unexpected endpoints.
Install Mechanism
Install spec uses a uv package (sift-gateway) which is a registry-based package installation — moderate-risk compared with direct downloads. This is proportionate for a skill that depends on that CLI, but users should confirm the package source and integrity before installing.
Credentials
The skill requests no environment variables or credentials. This is appropriate for a tool that operates via a local CLI and artifacts.
Persistence & Privilege
always:false (default) and autonomous invocation is the platform default. The skill does not request persistent system-wide privileges or modify other skills' configs.
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 reliable-tool-context - 安装完成后，直接呼叫该 Skill 的名称或使用
/reliable-tool-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 reliable-tool-context skill.
- Enables building reliable tool context from command output using artifacts.
- Facilitates reproducible and auditable querying with compact code snippets.
- Provides detailed decision rules for using Sift Gateway vs. direct CLI.
- Includes a schema discovery protocol and code query guardrails.
- Promotes disciplined, non-redundant model context and safe handling of paginated data.
元数据
常见问题
Reliable Tool Context 是什么？
Build reliable tool context from command output using artifacts and compact reproducible code queries. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 324 次。
如何安装 Reliable Tool Context？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install reliable-tool-context」即可一键安装，无需额外配置。
Reliable Tool Context 是免费的吗？
是的，Reliable Tool Context 完全免费（开源免费），可自由下载、安装和使用。
Reliable Tool Context 支持哪些平台？
Reliable Tool 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 Reliable Tool Context？
由 Lourenço Maciel（@zmaciel）开发并维护，当前版本 v1.0.0。
推荐 Skills