> ## Documentation index
> Fetch the complete documentation index at: https://contextrepo.com/docs/llms.txt
> Use it to find every page before exploring further.

# Frequently Asked Questions

> Answers to common questions about Context Repo — accounts, prompts, documents, MCP, API, and integrations.

Quick answers to the questions we hear most often. For deeper detail, follow the links to the relevant documentation pages.

## Account & Billing

### Is Context Repo free to use?

Every account starts with a 7-day free trial (on Pro or Pro Max) that includes the 29 hosted MCP tools, the Chrome Extension, version history, and document processing for 75+ file formats. After the trial, a paid plan (Pro, Pro Max, or Ultimate) keeps that access, with plan-specific monthly allowances for Reason, parsing, and page capture. See [Plans & Billing](https://contextrepo.com/docs/getting-started/plans-and-billing) for a full comparison.

### What do the higher plans add?

Every plan has the same dashboard, Deep Search, and tools. Pro Max doubles Pro's Reason and parse allowances and adds page capture through the Chrome Extension; Ultimate has ten times Pro's allowances. Visit the [pricing page](https://contextrepo.com/pricing) for current plan details.

## Prompts

### What are variables and how do I use them?

You can write placeholders such as `${variableName}` in prompt content, and the REST API can store variable metadata with string, number, or boolean types. Context Repo stores both verbatim; it does not detect placeholders, create fill-in fields, or substitute values. The client that retrieves the prompt handles any replacement. Learn more in the [Prompts guide](https://contextrepo.com/docs/dashboard/prompts).

### Does Context Repo track prompt changes?

Yes. Creation records version 0. Updates with non-empty prompt content, parameters, or variable metadata create another version; title, description, and tags-only updates do not. Restoring an earlier snapshot creates a new version instead of deleting later history. See [Prompts — Version History](https://contextrepo.com/docs/dashboard/prompts#version-history) for details.

### Can I share prompts publicly?

Open **Share** on a prompt and choose **Make public** to create a read-only link pinned to its current version. Anyone with the link can view that snapshot. Later edits stay private until you update the shared link, and **Stop sharing** revokes it.

## Documents

### What file formats can I upload?

Context Repo supports 75+ file formats including PDFs, Word documents, Excel spreadsheets, PowerPoints, images (with OCR), plain text, markdown, HTML, CSV, and many more. See the [Documents guide](https://contextrepo.com/docs/dashboard/documents) for the full list.

### Can I scrape web pages as documents?

Yes, on Pro Max and Ultimate. Enter a URL and Context Repo extracts the page content for you. You can configure options like extracting only the main content, including a screenshot, and setting a wait time for pages that load dynamically. Learn more in [Documents — Web Scraping](https://contextrepo.com/docs/dashboard/documents#web-scraping).

### How does Context Repo prevent document edits from overwriting each other?

Each completed document has a canonical revision. MCP `update_document` and
`restore_document_version` use the required `expectedRevision` from the latest
read or successful acknowledgement and return a conflict instead of
overwriting a newer change. Raw REST clients can also use
`expectedRevision` or `If-Match`. REST managed editing adds an active fenced
lease. The dashboard keeps your local draft when either check fails.

## MCP Server

### What is MCP?

The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external tools and data sources. Context Repo's hosted MCP Server registers 29 tools for prompts, documents, collections, search, retrieval, and cited reasoning. The inventory includes the OpenAI Apps SDK Company Knowledge tools `search` and `fetch`. Learn more in the [MCP overview](https://contextrepo.com/docs/mcp).

### What AI tools work with Context Repo?

Context Repo has documented setup paths for Claude, Cursor, ChatGPT, VS Code,
Windsurf, Factory, and Amp. The MCP ecosystem is large and growing, but each
client must support authenticated Streamable HTTP or OAuth, or be able to
launch the stdio bridge. Check the [setup guides](https://contextrepo.com/docs/mcp/setup-guides/cursor) for
popular clients or use the
[generic setup guide](https://contextrepo.com/docs/mcp/setup-guides/other-clients) to check those
requirements.

### How do I connect Context Repo to my AI assistant?

Each client has its own setup process. For Cursor, there's a [one-click deep link](https://contextrepo.com/docs/mcp/setup-guides/cursor). For Claude Desktop, you add a [JSON config file](https://contextrepo.com/docs/mcp/setup-guides/claude-desktop). For other clients, see the [setup guides](https://contextrepo.com/docs/mcp) for detailed instructions.

### What can I do with the 29 hosted MCP tools?

The tools are organized into 8 categories: User Info, Prompts (create, read, update, delete, version), Documents (create, read, update, delete, version), Collections (create, read, update, delete, manage items), Search (find items across your workspace), Deep Search (explore document content chunk by chunk), Reasoning (synthesized, cited answers across your documents), and Company Knowledge compatibility (`search` and `fetch`). See the [Tools Reference](https://contextrepo.com/docs/mcp/tools-reference) for full parameter details.

### When should I use Reasoning instead of Deep Search?

Use Deep Search when you want ranked passages to inspect and navigate. Use
Reasoning when you want one best-effort cited answer that reports limitations
it notices. Its `gaps` list is non-exhaustive; an empty list doesn't prove
complete coverage. Disagreements appear as ordinary cited statements in the
answer.
Reasoning is read-only and works on document content, not prompts. See the
[Reasoning guide](https://contextrepo.com/docs/concepts/reasoning) for examples.

## API

### How do I authenticate API requests?

The API supports two authentication methods: **Bearer JWT** (the session token issued to your signed-in browser) and **API-Key** (a key generated from Dashboard > Settings). Both use the `Authorization` header. See [Authentication](https://contextrepo.com/docs/api/authentication) for setup details.

### Where do I get an API key?

Go to **Dashboard > Settings > API Keys** and click **New API Key**. Choose a name, select permission scopes, and optionally set an expiration date. The full key is shown once — copy it immediately. See [Settings](https://contextrepo.com/docs/dashboard/settings) for details.

### What permission scopes are available for API keys?

Four scopes are available — one read and one write per resource: `prompts.read`, `prompts.write`, `documents.read`, and `documents.write`. The `.write` scopes already include the matching `.read` access on the same resource. The dashboard dialog presents this as **None / Read / Write** radio options for Prompts and Documents & Collections. See [Settings — Permission Scopes](https://contextrepo.com/docs/dashboard/settings#permission-scopes) for details.

### What endpoints are available?

The REST API includes endpoints for prompts, documents, collections, search, progressive disclosure, and reasoning. Every endpoint is documented with methods, paths, parameters, and examples in the [Endpoints reference](https://contextrepo.com/docs/api/endpoints).

## Integrations

### What does the Chrome Extension do?

The [Chrome Extension](https://contextrepo.com/docs/integrations/chrome-extension) uses a
`documents.write` API key and offers three paths: **Save this chat** for the
visible ChatGPT or Claude conversation, **Capture this page** for a public
HTTP(S) URL processed through Firecrawl, and **Pick element** for selected
content on a supported page. Direct reads can retrieve a document after the
write succeeds. Semantic and deep-search availability follows asynchronous
indexing.

### How does the ChatGPT App work?

The [ChatGPT App](https://contextrepo.com/docs/integrations/chatgpt-plugin) connects to your Context Repo account through OAuth and the same 29-tool hosted MCP server used by other clients. The inventory includes the OpenAI Apps SDK Company Knowledge tools `search` and `fetch`.

### How do I connect Context Repo to Grok?

Create a custom connector in Grok named **Context Repo** and set its server URL to `https://contextrepo.com/grok-mcp`. Grok will open Context Repo's OAuth sign-in flow automatically. You don't need an API key or client secret. Follow the [Grok integration guide](https://contextrepo.com/docs/integrations/grok) for the complete setup and verification steps.

### How do I connect Context Repo to Grok Bot?

Grok Bot is the desktop and iOS teammate app, not Grok chat on grok.com. In a Bot conversation, send `Add this MCP server: https://contextrepo.com/grok-mcp`. Grok Bot will confirm the details and open Context Repo's OAuth sign-in flow. You don't need an API key or client secret, and you don't add this from **Settings → Plugins**. Follow the [Grok Bot integration guide](https://contextrepo.com/docs/integrations/grok-bot) for the complete setup and verification steps.