osrs-wiki-mcp
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@osrs-wiki-mcpfind drop sources for abyssal whip"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
OSRS Wiki MCP
A stateless, local Model Context Protocol server that turns Old School RuneScape Wiki data into bounded semantic responses with validated structured content and source provenance.
Requirements and installation
Node.js 24 or newer is required. The executable prints a concise upgrade message on older supported runtimes.
Configure an MCP client to use the pinned npm release:
{
"command": "npx",
"args": ["-y", "osrs-wiki-mcp@1.0.0"]
}To run the server from a source checkout instead:
npm.cmd ci --ignore-scripts
npm.cmd run build
node dist/index.jsThe server speaks MCP over standard input/output. Standard output is reserved for protocol messages; sanitized startup diagnostics use standard error.
Related MCP server: OSRS MCP Server
Tools
All tools are read-only and may contact the Old School RuneScape Wiki.
Tool | Exact inputs | Purpose |
|
| Search for canonical pages and snippets. |
|
| Return a bounded cleaned page and its section summary. |
|
| List up to 200 page sections. |
|
| Return one bounded section using an index from |
|
| Normalize an item's description, properties, bonuses, and creation facts. |
|
| Return a complete paginated shop listing. |
|
| Return a complete paginated monster-drop listing. |
|
| Return a bounded overview of drops, shops, recipes, and ground spawns. |
|
| Return Wiki-sourced requirements without evaluating a player. |
|
| Return separate monster variants, map points, and access notes without DPS calculations. |
Public text inputs contain 1–256 Unicode characters. Truncation and incomplete
upstream pagination are disclosed in warnings; actionable warnings name a
follow-up tool or safe retry when one exists.
Responses and errors
Successful calls include both:
content: a readable representation for clients that display text; andstructuredContent: the same result validated against the tool's declared MCP output schema.
Valid calls that cannot complete return isError: true, no structured content,
and one stable code: NOT_FOUND, UPSTREAM_TIMEOUT,
UPSTREAM_RATE_LIMITED, UPSTREAM_UNAVAILABLE,
UPSTREAM_INVALID_RESPONSE, RESPONSE_TOO_LARGE, or INTERNAL_ERROR.
Malformed arguments and unknown tool names are JSON-RPC parameter errors.
Reliability and cache policy
Every tool call has a 30-second monotonic budget. Each logical upstream request has at most 20 seconds within the remaining tool budget.
Transient network errors and HTTP 429/502/503/504 responses receive at most two bounded retries.
Retry-Afteris honored only when it fits the remaining budgets.Upstream requests are serialized per process. Identical in-flight requests are deduplicated without allowing one caller's cancellation to cancel its siblings.
Responses are capped at 5 MiB after decompression. Bucket scans are capped at 20 pages of 500 raw rows (10,000 rows total).
Valid successful envelopes may be cached in memory for 300 seconds, up to 256 entries and approximately 32 MiB. Cache timestamps remain the original fetch time. Wiki error envelopes, invalid data, aborted requests, and oversized responses are never cached.
The cache is per-process only. Nothing is written to disk and restarting the process clears all state.
Provenance and licensing
Every successful response includes provenance with the contributing canonical
Wiki URLs, original fetchedAt, attribution party, license and license deed
URL, and a transformation indicator. Parsed-page responses additionally carry
the exact revision ID and revision URL. Bucket responses use fetch time when a
revision is not available without extra upstream requests.
The source code is MIT licensed. Content retrieved from the Old School RuneScape Wiki remains subject to CC BY-NC-SA 3.0, including its non-commercial and ShareAlike conditions. Downstream users are responsible for how they reuse that content. See THIRD_PARTY_NOTICES.md.
Testing
Offline tests use synthetic fixtures and never need the live Wiki:
npm.cmd ci --ignore-scripts
npm.cmd run typecheck
npm.cmd test
npm.cmd run build
npm.cmd run smoke:stdio
npm.cmd run pack:checkThe live smoke test is opt-in. It builds the server and makes exactly two
low-volume requests (search_wiki and get_item_info); it never stores the
response:
npm.cmd run test:liveScope and limitations
Version 1 deliberately excludes player accounts, progression evaluation, hiscores, Grand Exchange prices, DPS calculations, Wiki images, persistent storage, a hosted proxy, and a local Wiki mirror. Quest requirements are facts, not met/missing evaluations. Monster variants are never combined.
RuneScape and Old School RuneScape are trademarks of Jagex Limited. This independent project is not affiliated with, endorsed by, or sponsored by Jagex Limited, Weird Gloop, or the Old School RuneScape Wiki. It uses a descriptive User-Agent with this repository URL and is designed for distributed, low-volume, per-user access.
See CONTRIBUTING.md for development rules and SECURITY.md for private vulnerability reporting.
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Latest Blog Posts
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/SanderVirula/osrs-wiki-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server