Guide · Agent skill

How to use the External Context Search agent skill

Updated 2026-10-29 · written with AI from the project's own documentation and checked against our catalogue data

The External Context Search skill (external-context) is an agent skill for Claude that answers questions about a technology or topic by searching the web and reading documentation in parallel. It is meant for people who need a source-linked overview rather than an answer from the model's memory. After this guide you will know what the skill does, how to install it, how to trigger it with a command or a natural request, and what to check before relying on its results.

External Context SearchYeachan-Heo/oh-my-claudecode · MIT Download the skill

Quick start

1Install the oh-my-claudecode plugin. The external-context skill and the document-specialist agent both come from this plugin. Install it before using the skill.
2Download the skill ZIP. Get the external-context ZIP from this page.
3Unzip into a skills folder. Extract it into ~/.claude/skills/ for personal use or .claude/skills/ inside a project.
4Or upload in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the ZIP.
5Invoke the skill explicitly. Run /oh-my-claudecode:external-context followed by your question. Natural requests alone do not trigger it.
6Check the sources. Open the linked citations and confirm the key findings before relying on them.

What is an agent skill and what does this one do?

An agent skill is a set of instructions packaged as a folder that Claude can load when a task matches it. This skill is defined in a single SKILL.md file, which describes how the agent should handle external research requests.

When you ask a question about a technology or topic, the skill tells the agent to split the question into 2 to 5 separate search angles. Each angle goes to a helper agent called document-specialist, which searches the web and reads documentation at the same time. The main agent then combines the results into one summary with key findings and links to the sources it used.

It is most useful when the answer depends on current or official documentation rather than what the model already knows.

Splits a query into 2 to 5 independent facets
Runs document-specialist agents in parallel, with a maximum of 5
Returns key findings, detailed results per facet and a list of sources with URLs

How do you install the External Context skill?

The skill is part of the oh-my-claudecode plugin, which provides both the external-context skill and the document-specialist agent. Download the skill ZIP from this page and use one of the two methods below.

For Claude Code, unzip the download into your personal skills folder ~/.claude/skills/ or into a project-level .claude/skills/ folder inside your project. Project-level installation keeps the skill available only to that project.

In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

Personal Claude Code install: unzip into ~/.claude/skills/
Project install: unzip into .claude/skills/ in your project
Claude apps: Settings → Capabilities → Skills, then upload the ZIP

What do you need before you run it?

The skill depends on the oh-my-claudecode plugin, because that plugin supplies the document-specialist agent that does the searching. You also need Claude Code with web search and web fetch tools available to the agents. Without those tools the helper agents cannot look anything up.

This skill runs no scripts, but it does use the network. Queries and the pages the agents fetch are sent to external web services.

oh-my-claudecode plugin installed
Claude Code with WebSearch and WebFetch tools available to agents
Network access (the skill uses the network; it runs no scripts)

How do you trigger the skill?

You can invoke the skill explicitly with a slash command followed by your topic or question. The argument hint for the skill is a search query or topic.

The skill has no magic keyword trigger, so it only runs when you invoke it explicitly. A natural request alone will not start it, so use the command form when you want this behaviour.

Explicit command: /oh-my-claudecode:external-context <topic or question>
No automatic keyword trigger: it runs only when invoked explicitly

/oh-my-claudecode:external-context <topic or question>

What are good examples of queries?

Specific questions work best, especially ones that name a technology and a goal. Comparison questions are useful because each tool can become its own search facet.

/oh-my-claudecode:external-context What are the best practices for JWT token rotation in Node.js?
/oh-my-claudecode:external-context Compare Prisma vs Drizzle ORM for PostgreSQL
/oh-my-claudecode:external-context Latest React Server Components patterns and conventions

What does the output look like?

The result starts with the original query and a numbered list of key findings, each with a link to its source. Below that, detailed results are grouped by facet with citations, followed by a full list of sources.

Because the skill relies on whatever pages the helper agents find, check the citations before acting on them. The agents can still misread or miss sources.

Key Findings: numbered, each linked to a source
Detailed Results: one section per facet with citations
Sources: list of URLs used

What should you watch out for?

Do not send private or confidential data through this skill, since queries and fetched pages go to external web services. Do not treat its output as fully verified; open the linked sources yourself when the answer matters.

The skill is not a replacement for checking official documentation when the sources must be exact.

Do not use it for private or confidential information
Verify citations before acting on them
Parallel agents mean several web lookups run per request

What people use it for

Checking current best practices for a library or framework
Comparing two tools, such as two ORMs, with cited sources
Getting a quick, source-linked overview of a topic for study or analysis
Finding official documentation links to share with a team

Frequently asked questions

How do I install the external-context skill?

Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into .claude/skills/ in a project. In the Claude apps you can upload the ZIP under Settings → Capabilities → Skills.

Is the external-context skill free?

The source material for this guide does not state pricing or licence terms beyond the item's MIT licence, so check the plugin's own listing for any cost details. The skill is listed under the MIT licence.

What does the external-context skill require?

It requires the oh-my-claudecode plugin, which provides the document-specialist agent, and Claude Code with web search and web fetch tools available to the agents. It uses the network and runs no scripts.

How do I trigger the external-context skill?

Invoke it explicitly with /oh-my-claudecode:external-context followed by your topic or question. It has no magic keyword trigger, so a plain request will not start it.

Is it safe to use with private data?

No. Queries and fetched pages are sent to external web services, so keep private or confidential information out of it.

Can I trust the citations it gives?

Not without checking. The skill relies on whatever pages the helper agents find and can misread or miss sources, so open the links and verify them before acting on the results.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use this skill if you are a developer, student or analyst who needs a quick, source-linked overview of current or official documentation, or a cited comparison of tools. Pick something else if your question involves private data or you need fully verified sources, and do not rely on its citations without checking them.

Download the skill