Token导航 LogoToken导航TokenDH.com
研究检索操作浏览器github未标认证来源可访问clear审计异常

forum-research论坛研究

Agent Skill

forum-research 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

324

周安装

13

GitHub Stars

9

下载量

105
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

3

许可证

MIT

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:forum-research(论坛研究)
来源仓库:https://github.com/jorgenbuilder/icp-skills
仓库路径:skills/forum-research
安装命令:
npx skills add https://github.com/jorgenbuilder/icp-skills --skill forum-research
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。不同来源提供的安装方式可能略有差异;本站展示可直接复制的安装命令,安装前请核对来源页面。

skills.shnpx skills
npx skills add https://github.com/jorgenbuilder/icp-skills --skill forum-research

简介

forum-research 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中快速定位候选结果。

  • 适用于论坛或社区讨论相关的信息检索任务。
  • 可通过来源仓库和 README 文档核验具体功能和使用方法。
  • 安装前应确认权限、维护状态及是否涉及网络或文件操作。
  • 建议结合论坛平台特性验证其搜索能力。

SKILL.md

Forum Research (Dfinity)

Scope

This skill supports read-only research on forum.dfinity.org using authenticated browsing and Discourse JSON endpoints.

Required dependency

This skill depends on the agent-browser skill from vercel-labs. Install it first:

npx skills add https://github.com/vercel-labs/agent-browser --skill agent-browser

Guardrails (strict)

  • Read-only only. Never post, reply, like, bookmark, or edit.
  • Use GET requests for all research and reading.
  • Exception: authentication may require a POST to the login endpoint. No other non-GET requests are allowed.
  • If a user asks to post or modify content, refuse and explain the read-only policy.

Browser initialization (CRITICAL)

ALWAYS call agent-browser launch first before ANY other agent-browser commands.

If you get the error "Browser not launched. Call launch first", simply call agent-browser launch and then retry your command. This is a normal part of the workflow.

Authentication (CRITICAL)

You MUST be logged in to access forum content. Most forum content is hidden without authentication.

REQUIRED workflow for every session:

  1. Launch browser with agent-browser launch
  2. Login IMMEDIATELY before doing any research (see Stage 1 below)
  3. Verify login by checking the top right of any forum page:

- Logged in: Account dropdown visible in top right - Not logged in: "Sign Up" and "Log In" buttons visible in top right

  1. Only proceed with research tasks after confirming you're logged in

During long sessions:

  • Periodically check the top right of the page to verify you're still logged in
  • If you see "Sign Up" and "Log In" buttons instead of account dropdown, re-authenticate immediately
  • Sessions may expire after inactivity - re-login if needed

Error handling and persistence

When using this skill, you may encounter errors. Always persist through errors by:

  1. Launch errors: If you see "Browser not launched", call agent-browser launch and retry.
  2. Login errors: If credentials fail, try again or check the credential source.
  3. Access denied / empty results: You're not logged in - complete the login process immediately.
  4. Navigation errors: If a page fails to load, retry the navigation.
  5. Element not found: If snapshot elements change, take a fresh snapshot and update element refs.

Do not give up after a single error. Most errors can be resolved by:

  • Launching the browser if not already launched
  • Ensuring you're logged in (check top right of page for account dropdown vs login buttons)
  • Retrying the failed command
  • Taking a fresh snapshot to update element references
  • Verifying you're on the expected page before interacting with it

Stage 1: Login (MANDATORY FIRST STEP)

This MUST be the first thing you do in every session. Do not skip this step.

  1. Launch the browser with agent-browser launch (required first step).
  2. Obtain credentials using one of the approved methods in reference.md.
  3. Navigate to https://forum.dfinity.org/login.
  4. Always choose the username/password option on the login page (ignore GitHub and passkey options).
  5. Fill username and password, then submit.
  6. Confirm login by taking a snapshot and checking the top right of the page:

- Look for an account dropdown (logged in successfully) - If you see "Sign Up" and "Log In" buttons instead, login failed - retry the login process

  1. Alternative verification: Open https://forum.dfinity.org/u/<username>.json - you should see your user profile data.

After login confirmation, you can proceed with research tasks.

Discourse JSON access

Discourse supports a JSON view for most pages by appending .json to the URL. Prefer JSON for structured reading:

  • Topic: https://forum.dfinity.org/t/<slug>/<id>.json
  • User: https://forum.dfinity.org/u/<username>.json
  • Latest: https://forum.dfinity.org/latest.json

Additional resources

  • Credential sourcing, prompts, and constraints: reference.md
  • Login examples: examples.md
  • Forum category structure and navigation guide: categories.md

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

04

需要参考平台分布和安装热度时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Claude Code

73.45%
按下载量换算77

安全审计

Gen Agent Trust Hub

可疑

Socket

通过

Snyk

未通过

权限和风险

操作浏览器

该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。

来源信息

继续浏览同类 Skills