RimSearcher, developed by Kearril, is an MCP server that connects LLM-based coding assistants to a local RimWorld codebase to provide direct file context and search capabilities. It indexes C# and XML game files and exposes six MCP tools for locating, inspecting, tracing, reading code, plus regex search and directory listing, so the assistant can generate grounded suggestions and debug logic. The tool targets mod developers who use AI assistants and need accurate, local code context for modding workflows.

Supplies assistants with resolved project symbols and Def data

RimSearcher builds a local indexed database of on-disk C# and XML and makes that project state available through MCP endpoints. The implementation uses Roslyn-based C# parsing for syntactic inspection and a custom XML resolver that evaluates ParentName and Name attributes so a model can reference fully resolved Defs. That query surface reduces manual file hunting by letting an assistant fetch precise declarations and assembled Def data.

Improves grounding in generated suggestions but not a substitute for review

By providing direct, indexed references and tracking cross-references between classes and Defs the tool reduces the chance an assistant invents out-of-repo symbols. The XML inheritance resolution and Roslyn analysis make returned snippets verifiable against the repository; however generated code and logic still require human validation and in-game testing for runtime correctness and gameplay effects.

Operates on local RimWorld files and requires an MCP client

RimSearcher runs against a local RimWorld installation and configurable mod directories, processing C# source and XML Def files. It functions as an MCP server and expects an MCP-compatible client such as Claude Desktop and a .NET-capable environment. Indexing uses an SQLite-backed store designed for concurrency, so large mod collections can be indexed and queried on the developer's machine.

Suits active modders but assumes technical setup and maintenance

The tool's local execution model preserves repository privacy and supports offline core functions, which aligns with private mod development. Index updates and permissioning are part of normal use, so the workflow favors developers comfortable configuring an MCP client, running index refreshes after code changes, and integrating assistant outputs into standard review and testing cycles rather than casual experimentation.

Practical for modders who want project-aware AI, with verification required

RimSearcher suits mod developers who need AI assistance tied to their current project state; expect to treat generated suggestions as starting points and verify them through local testing and code review. Pair the tool with version control and in-game testing as part of an established development workflow. It fits collaborative teams and solo maintainers aiming to reduce manual code lookup while preserving hands-on verification.