Use case · Free plugin

# My agent read a whole upgrade guide to find two sections. *Now it reads the sections it needs.*

When your agent reads a docs page, a web page or a browser snapshot, the free context-mode plugin puts the page in a local index. The agent reads the sections that answer the question. A fetch of the 265 KB Vite features guide reached the model as 0.4 KB.

[Install the free plugin](https://contextmode.ai/docs/engine/install) [Read the docs](https://contextmode.ai/docs/engine/search)

How do I set up the dev server proxy in Vite?

Full page Free plugin

Full page: navigation, tables and every example reach the model.

WebFetch returns **265.4 KB** The model reads **265.4 KB**

Every later call sends it again.

Free plugin: the page goes to a local index, and the agent searches it.

The index holds **265.4 KB** The model reads **0.4 KB**

Sent to the model

- The proxy section, with its source

Kept on your machine

- The full page, searchable with ctx_search

The problem

## The answer is two sections. *The page is 265 KB.*

**Each page is long**

Docs pages hold navigation, examples for other setups and long tables.

**The agent needs a few parts**

An upgrade guide, an API reference and two blog posts: a few sections from each.

**Every page stays in the thread**

Text in the conversation goes back to the model on every later call of the session.

How it works

## The page goes to an index. *The agent reads the answer.*

1. 01 **The hook catches the fetch** The PreToolUse hook sees a large WebFetch or Read and sends it to the sandbox.
2. 02 **The page goes into the index** A SQLite FTS5 index (SQLite full-text search) in `~/.context-mode`, one database per project.
3. 03 **The agent searches it** It calls `ctx_search` with its question and reads the best sections, with their source. A second question costs one search.

Proof

## Four sources, measured. *The model reads under 1 KB of each.*

| Source | Full size | The model read |
| --- | --- | --- |
| WebFetch of the Vite features guide (the hook sends it to `ctx_index` and `ctx_search` ) | 265.4 KB | 0.4 KB |
| Browser snapshot ( `take_snapshot` , the Wikipedia SQLite page) | 133.3 KB | 0.5 KB |
| Read of the 89 KB Vite config source (the hook sends it to `ctx_execute` ) | 89.2 KB | 0.4 KB |
| Library docs ( `query-docs` , Vite build and proxy config) | 6.8 KB | 0.5 KB |

Plugin benchmark of 2026-11-01 on an Apple M1 Pro, default settings, median of 5 runs. Raw data: [the result tables](https://contextmode.ai/benchmarks/engine/2026-11-01-usecases.md). Rerun it with the steps on [Benchmarks](https://contextmode.ai/docs/engine/benchmarks).

Prompts

## Give your agent *one of these.*

Prompt

```
Read the Vite docs on build options and the dev server proxy. Write the proxy setup for an API that runs on port 8080, and cite the section you used.
```

Prompt

```
Read the React 19 upgrade guide. List each change that touches this repo, with the file to edit, then make the changes.
```

Prompt

```
Compare the Stripe and Paddle webhook docs. Write a handler for each that checks the signature, and list what differs.
```

Products used

## What this *uses.*

**[The free plugin](https://contextmode.ai/engine)**

Open source. It runs in your agent, on your machine.

**[Search and index](https://contextmode.ai/docs/engine/search)**

How the page goes into the index and how the agent finds a section.

**[How it works](https://contextmode.ai/docs/engine/how-it-works)**

The hooks, the sandbox and the local index.

**Related**

[Free plugin resources](https://contextmode.ai/open-source/resources) · [Test and build logs](https://contextmode.ai/use-cases/test-logs) · [Analyze large files](https://contextmode.ai/use-cases/analyze-large-files)

## Index the whole page. *Read the part you need.*

Each agent installs the free plugin with its own install command.

[Install the free plugin →](https://contextmode.ai/docs/engine/install) [Source on GitHub →](https://github.com/mksglu/context-mode)