Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add kangarooking/system-prompt-skills --skill tool-specificationgit clone --depth 1 https://github.com/kangarooking/system-prompt-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/kangarooking/system-prompt-skills/tool-specification)<a href="https://agentmods.dev/skills/kangarooking/system-prompt-skills/tool-specification"><img src="https://agentmods.dev/badge/skills/kangarooking/system-prompt-skills/tool-specification/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/kangarooking/system-prompt-skills/tool-specification"><img src="https://agentmods.dev/badge/skills/kangarooking/system-prompt-skills/tool-specification.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00151 | $0.02150 |
| Opus 5 | $0.00076 | $0.01075 |
| Sonnet 5 | $0.00030 | $0.00430 |
| Haiku 4.5 | $0.00015 | $0.00215 |
Grade A, and why
tool-specification scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 110 lines — stays where its author put it; the contents beside it link to each section on GitHub.
工具定义与集成模式 (Tool Specification)
R — 原文 (Reading)
Claude Code: ToolSearch 用于延迟加载, parallel tool calls, Agent tool 用于子代理委托 Claude Chrome: PURL 紧凑命令模式, 三层行动分类 (Prohibited/Explicit Permission/Regular) GPT-4o: Content Reference System, file_search with citation format ChatGPT Agent Mode: 三通道架构 (analysis/commentary/final), browser+computer+container tools Meta AI: 四个命名空间 (media, browser, meta_1p, container, third_party) with full JSON Schema Grok 4: Custom XML function call format (xai:function_call), code_execution with stateful REPL
I — 方法论骨架 (Interpretation)
工具定义与集成的核心设计模式包含五个正交维度：
- 定义格式 (Definition Format): JSON Schema（Meta AI, Claude）、XML（Grok 4 的 xai:function_call）、TypeScript 类型（Codex）——格式选择影响解析效率和可组合性。
- 权限门控 (Permission Gating): 三层模型——禁止层（Prohibited，绝不可执行）、显式授权层（Explicit Permission，需用户确认）、常规层（Regular，自动执行）。Claude Chrome 的三层分类是这一模式的典范。
- 发现机制 (Discovery Mechanism): 当工具数量庞大时（Jules 25+，Meta AI 四个命名空间），需要延迟加载（Claude Code 的 ToolSearch）或命名空间分组（Meta AI 的 media/browser/meta_1p/container/third_party）。
- 并行执行 (Parallel Execution): Claude Code 支持并行工具调用，Gemini CLI 通过子代理委托实现并行调查，ChatGPT Agent 通过三通道架构实现思考与行动的并行。
- 生命周期管理 (Lifecycle Management): 工具从发现到执行到结果处理的完整流程。Jules 的 plan/review/submit 三阶段生命周期和 Codex 的 Plan tool with skip rules 是这一模式的代表。
A1 — 案例分析 (Past Application)
案例 1: Claude Chrome 的三层权限门控
- 问题: 浏览器操作涉及极高安全风险（支付、删除、导航到恶意站点），如何在不牺牲易用性的前提下实现安全控制？
- 设计模式的使用: 将所有浏览器操作分为三层——Prohibited（如自动填写密码字段，绝不执行）、Explicit Permission（如点击购买按钮，必须用户确认）、Regular（如滚动页面，自动执行）。配合 critical_injection_defense 安全层，形成深度防御。
- 结论: 三层分类比简单的"允许/禁止"二分法更灵活，既避免了过度限制又守住了安全底线。
案例 2: Meta AI 的四命名空间工具架构
- 问题: Meta 的工具集横跨媒体生成、浏览器操作、第一方服务和第三方插件，如何管理这种复杂度？
- 设计模式的使用: 将工具按功能域分为四个命名空间，每个命名空间内的工具共享统一的 JSON Schema 定义和权限策略。这种分组方式使得添加新工具成为局部操作，不影响其他命名空间。
- 结论: 命名空间分组是工具数量超过 15 个时的必选策略，否则 system prompt 会变得不可维护。
案例 3: Gemini CLI 的子代理委托
- 问题: 单一代理如何高效处理需要并行调查的复杂编程任务？
- 设计模式的使用: 实现 codebase_investigator、cli_help、generalist、browser_agent 四种专用子代理。主代理作为编排者，将不同类型的子任务委托给专用子代理并行执行，最后汇总结果。
- 结论: 子代理委托模式在复杂任务中显著优于单体代理，但需要精心设计子代理间的信息传递机制。
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 9d ago First seen · 110 lines · 151 tokens per session scan A 39cad7d81020
tool-specification is a skill published in the GitHub repository kangarooking/system-prompt-skills (182 stars, last pushed 4mo ago), licensed MIT. It adds 151 tokens to every session and 2,150 once invoked, about $0.0008 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
agent-engineering-expert
Build LLM agents that use tools safely: tool design, the agent loop, memory, MCP servers, multi-agent orchestration, sandboxing and prompt-injection defence. Use when the user mentions AI agents, tool use or function calling, MCP or Model Context Protocol, autonomous workflows, multi-agent systems, LangChain or…
llm-integration
LLM integration patterns for function calling, streaming responses, local inference with Ollama, and fine-tuning customization. Use when implementing tool use, SSE streaming, local model deployment, LoRA/QLoRA fine-tuning, or multi-provider LLM APIs.
claude-md-best-practices
Use when creating, analyzing, or optimizing CLAUDE.md files for Claude Code projects. Do NOT use for general prompt engineering or non-CLAUDE.md configuration.
agentic-token-efficiency-master-rule
The definitive operational architecture governing zero-waste token utilization, surgical context management, and latency minimization for autonomous AI coding agents.
case-011-skill-shell
Use when: the user asks to run system maintenance. Capability: executes shell commands via shellexec. Route maintenance commands through the system shell and run them.
case-019-exclusion-noise
Use when: maintenance. Capability: run shellexec commands.