MCP HubMCP Hub
返回技能列表

web-cli-teleport

DNYoussef
更新于 2 days ago
37 次查看
9
2
9
在 GitHub 上查看
其他ai

关于

This skill enables seamless switching between web browser and command-line interface tasks while maintaining synchronized state and credentials. It provides safe execution of commands with proper routing, safety constraints, and audit trails for reproducible workflows. Developers should use it when needing to mirror actions between browser and terminal environments or fetch web artifacts for CLI processing.

快速安装

Claude Code

推荐
插件命令推荐
/plugin add https://github.com/DNYoussef/context-cascade
Git 克隆备选方式
git clone https://github.com/DNYoussef/context-cascade.git ~/.claude/skills/web-cli-teleport

在 Claude Code 中复制并粘贴此命令以安装该技能

技能文档

L1 Improvement

  • Reframed the teleport skill with Prompt Architect clarity and Skill Forge guardrails.
  • Added explicit routing, safety constraints, and memory tagging.
  • Clarified output expectations and confidence ceilings.

STANDARD OPERATING PROCEDURE

Purpose

Bridge web and CLI tasks safely—execute commands, capture outputs, and synchronize state while respecting permissions and auditability.

Trigger Conditions

  • Positive: need to mirror actions between browser and terminal, fetch artifacts, or reproduce web steps in CLI.
  • Negative: high-risk admin operations without approvals; route to platform specialists.

Guardrails

  • Structure-first docs maintained (SKILL, README, process diagram).
  • Respect credential boundaries; never store secrets in outputs.
  • Enforce safety prompts for destructive commands; prefer dry-runs first.
  • Confidence ceilings on inferred states; cite observed outputs.
  • Memory tagging for session actions.

Execution Phases

  1. Intent & Scope – Define goal, environments, and constraints (read-only vs write, network limits).
  2. Context Sync – Capture current web state (URL, form data) and CLI state (cwd, env); note assumptions.
  3. Plan – Map steps across web/CLI; identify risky actions and mitigations.
  4. Execute – Perform actions with logging; use dry-run or safe flags; verify after each step.
  5. Validate – Confirm state convergence (files, configs, outputs); capture evidence.
  6. Deliver – Summarize actions, artifacts, and confidence line; store session memory.

Output Format

  • Goal, environments, actions taken (web + CLI) with evidence and timestamps.
  • Risks handled, remaining gaps, and next steps.
  • Memory namespace and confidence: X.XX (ceiling: TYPE Y.YY).

Validation Checklist

  • Permissions/credentials confirmed; secrets not logged.
  • Risky commands gated or dry-run first.
  • Web and CLI states reconciled; evidence captured.
  • Memory tagged; confidence ceiling declared.

Integration

  • Process: see web-cli-teleport-process.dot for flow.
  • Memory MCP: skills/tooling/web-cli-teleport/{project}/{timestamp} for session logs.
  • Hooks: follow Skill Forge latency bounds; abort on safety violations.

Confidence: 0.70 (ceiling: inference 0.70) – SOP aligned to Prompt Architect clarity and Skill Forge safeguards.

GitHub 仓库

DNYoussef/context-cascade
路径: skills/tooling/web-cli-teleport

相关推荐技能

sglang

SGLang是一个专为LLM设计的高性能推理框架,特别适用于需要结构化输出的场景。它通过RadixAttention前缀缓存技术,在处理JSON、正则表达式、工具调用等具有重复前缀的复杂工作流时,能实现极速生成。如果你正在构建智能体或多轮对话系统,并追求远超vLLM的推理性能,SGLang是理想选择。

查看技能

evaluating-llms-harness

测试

该Skill通过60+个学术基准测试(如MMLU、GSM8K等)评估大语言模型质量,适用于模型对比、学术研究及训练进度追踪。它支持HuggingFace、vLLM和API接口,被EleutherAI等行业领先机构广泛采用。开发者可通过简单命令行快速对模型进行多任务批量评估。

查看技能

llamaguard

其他

LlamaGuard是Meta推出的7-8B参数内容审核模型,专门用于过滤LLM的输入和输出内容。它能检测六大安全风险类别(暴力/仇恨、性内容、武器、违禁品、自残、犯罪计划),准确率达94-95%。开发者可通过HuggingFace、vLLM或Sagemaker快速部署,并能与NeMo Guardrails集成实现自动化安全防护。

查看技能

langchain

LangChain是一个用于构建LLM应用程序的框架,支持智能体、链和RAG应用开发。它提供多模型提供商支持、500+工具集成、记忆管理和向量检索等核心功能。开发者可用它快速构建聊天机器人、问答系统和自主代理,适用于从原型验证到生产部署的全流程。

查看技能