Developer tools
Semlith
Local search over your code and documents for AI agents
- Category
- Developer tools
- Stage
- Beta
- For
- Developers and teams using AI coding agents who need searchable context across code repositories, documents, and project tools.
About
One index of everything you work on, for every agent you run. Free local MCP search over code, PDFs and Office files for Claude Code, Codex and Cursor.
Features
- Problem
- Agents start with incomplete context across repositories and work environments.
- How it works
- Indexes selected sources once and makes the index available to supported agents over MCP, locally or through Semlith Cloud.
- Result
- Agents can search the same selected corpus across machines and sessions.
- Why
- A shared index can serve agents in terminals, editors, cloud sessions, and CI without rebuilding context for each one.
- Problem
- Relevant answers may be in documents and project tools, not just source code.
- How it works
- Searches code alongside supported document formats, images, GitHub, GitLab, Confluence, Jira, folders, and uploads.
- Result
- Results can bring together code, pages, tickets, and files.
- Why
- Combining sources lets a query find related context wherever it is stored.
- Problem
- Text matching alone can miss relevant code or document passages.
- How it works
- Combines keyword and meaning-based search and maps code relationships across repositories.
- Result
- Agents can retrieve relevant passages and inspect connected code.
- Why
- Both conceptual queries and exact names matter, while graph relationships help answer callers and change-impact questions.
- Problem
- Retrieval systems that generate answers can introduce unsupported claims.
- How it works
- Retrieves indexed facts without a generative model in the retrieval path; answers identify the commit read.
- Result
- Agents receive source-grounded results rather than model-generated retrieval claims.
- Why
- Keeping retrieval separate from generation makes the source and freshness of retrieved context inspectable.
- Problem
- Teams need to understand what agents accessed and what code changes may affect.
- How it works
- Records retrievals in a hash-chained ledger and can post GitHub pull-request impact checks with callers and related documents.
- Result
- Teams can inspect retrieval history and review impact findings tied to indexed files.
- Why
- An auditable record and explicit graph findings help teams review agent access and change reach.
Product questions
What does Semlith do?
Semlith builds one searchable index of code, documents, tickets, and other project context for AI agents. It runs locally as an open-source binary or as an optional hosted service, and agents access the index over MCP.
Who is Semlith for?
Developers and teams using AI coding agents who need searchable context across code repositories, documents, and project tools.
Screenshots
