Your agents run in a lot of places now. A terminal, an editor, a cloud session, a CI job. Each one starts from zero inside one repository, so it greps and reads whole files to find what it needs. The caller lives in another repo, the spec is in Confluence, the decision is in a Jira ticket and the diagram is in a PDF.
Semlith indexes your repositories and documents once, keeps the index fresh as files change, and answers any agent over MCP. Claude Code, Codex, Cursor and other MCP clients get back the exact file and line instead of reading whole files. It reads PDF, Word, PowerPoint and Excel too, keeps a code graph for callers and impact, and records every retrieval so you can see what each agent read.
On SWE-bench Verified, the file the fix edits is in Semlith's top 10 for 76.8% of 500 issues, against 47.0% for grep then read. Retrieval only.
Semlith runs on your machine and is free. Semlith Cloud is the paid, hosted version. It keeps one index that every agent can reach, on any laptop, in the cloud or in CI, for you alone or your whole team.
The binary is open source (Apache-2.0), has every feature and never sends anything off your machine. It runs on macOS (Apple silicon), Linux (x86_64, ARM) and Windows (x86_64), and sets itself up for 12 tested agent clients. Semlith Cloud reads GitHub, GitLab, Confluence, Jira and uploads, read only. Pro starts at $5 a month and Team at $10 a seat a month, both founding prices. Pro has a 14-day trial with no card, and Team is by invitation while Cloud opens. Queries are never metered.
No comments or reviews, maybe you want to be first?