About OCI Documentation MCP Server
OCI Documentation MCP Server lets agents search and read OCI documentation. It combines Oracle Help Center search with paged Markdown access to individual documentation pages.
Key features
- 01Search OCI documentation
- 02Convert documentation pages to Markdown
- 03Read content by line window
- 04Return page metadata and table of contents
Use cases
- 01Find documentation for an OCI service or error
- 02Read selected sections of a long page
- 03Navigate documentation using headings and line numbers
MCP support
The server exposes `oci_search_documentation` and `oci_read_documentation` tools to MCP clients. It supports stdio, SSE, and streamable HTTP transports.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
uvx oci-documentation-mcp-server@latestFASTMCP_LOG_LEVELLocal servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEInspired by: https://github.com/awslabs/mcp/tree/main/src/aws-documentation-mcp-server
OCI Documentation MCP Server
Model Context Protocol (MCP) server for OCI Documentation
This MCP server provides tools to search for content, and access OCI documentation.
Change log
- 2026-05-20: support transport: stdio,sse,streamable-http
- 2026-05-19: change search engine to oracle help center search
- 2025-04-21: Initial release
Features
oci_search_documentation
Searches OCI documentation through the Oracle Help Center Search API and returns structured page results. This tool is intended for the first step of a documentation workflow: finding the most relevant Oracle documentation URL before reading the page.
Parameters:
search_phrase: Search text. Use specific OCI service names, product terms, error messages, or feature names for better results.limit: Maximum number of results to return. Defaults to3.page: Search result page number. Defaults to1.
Returns:
- Pagination metadata from the Oracle Help Center result set.
- A list of documentation results with title, URL, and description.
Design notes:
- Uses the public Oracle Help Center pages endpoint.
oci_read_documentation
Reads one OCI documentation page, converts it from HTML to Markdown, indexes it by line number, and returns a window of content. This tool is intended for controlled reading of long documentation pages without flooding the MCP response.
Parameters:
url: OCI documentation page URL. The URL must be fromdocs.oracle.comand must end with.htmor.html.start_index: 0-based line number to start reading from. Defaults to0.max_lines: Maximum number of Markdown lines to return. Defaults to10.
Returns:
stats: Total lines, total words, start line, returned lines, remaining lines, and remaining words.content: Markdown text for the requested line window.table_of_contents: Returned only whenstart_index == 0; includes heading level, title, and 0-based line number.
Design notes:
- Long documents are paged by Markdown line number rather than character offset, which makes follow-up reads easier for agents.
- Converted pages are cached in process memory for 24 hours, up to 128 pages. The cache stores a single line-list representation to avoid duplicating full Markdown text and split lines.
- Table of contents and related links are returned only for the first read to avoid repeating metadata during follow-up reads.
Use
Option 1: Run from pypi package
Defalt output through stdio, change that use --transport if you want.
{
"mcpServers": {
"oci-documentation-mcp-server": {
"command": "uvx",
"args": [
"--from",
"oci-documentation-mcp-server@latest",
"python",
"-m",
"oci_documentation_mcp_server.server",
"--transport",
"stdio"
],
"env": {
"FASTMCP_LOG_LEVEL": "ERROR"
}
}
}
}
Option 2: Run locally from source code and output through stdio
Installation Requirements
- Doenload this repo.
2.Install
uvfrom Astral or the GitHub README
{
"mcpServers": {
"oci-documentation-mcp-server": {
"command": "uv",
"args": [
"--directory",
"path/to/oci-documentation-mcp-server"
"run",
"python",
"-m",
"oci_documentation_mcp_server.server",
"--transport",
"stdio"
],
"env": {
"FASTMCP_LOG_LEVEL": "ERROR"
}
}
}
}
Option 3: Run as server
Run as server use Streamable HTTP:
uv run python -m oci_documentation_mcp_server.server --transport "streamable-http" --port 8000 --path "/mcp"
Config on agent tools:
{
"mcpServers": {
"oci-documentation-mcp-server": {
"type": "streamable-http",
"url": "http://localhost:8000/mcp"
}
}
}
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.