Reported News
Kensaku: a free, local file launcher that searches inside documents and screenshots, with MCP for agents
Kensaku is an MIT-licensed file launcher for Windows, macOS and Linux, created on GitHub on Oct 24, 2026, whose README says "You don't remember the filename. You remember what was in it." It searches text, OCR'd screenshots and meaning locally, exposes the index over MCP, and had 403 GitHub stars on Nov 1, 2026.
- Changes reported
- 5
- listed below, each from a source we read
- We tested
- 0
- we did not run it ourselves
- Gaps we flag
- 3
- things we could not verify
- Sources
- 2
- 1 primary, linked below
What changed
| # | What the source says | Status |
|---|---|---|
| 01 | Kensaku is a desktop launcher (summon with a keystroke, type, press Enter) that searches file names, text inside files, typo-tolerant matches and meaning in one search. The README lists PDF, DOCX, PPTX, XLSX, EPUB, ODT/ODS/ODP, RTF, plain text, code and logs, including CJK text. | Reported |
| 02 | Text in images and screenshots is read with PP-OCRv6 by default, with an optional Indra engine (about a 27 MB pack downloaded on demand) or Tesseract. Queries can use natural-language time such as "last week" and a query language with filters like type:, size:, modified: and re:. | Reported |
| 03 | For agents, MCP clients point at kensaku-cli mcp; kensaku-cli serve offers loopback-only HTTP with a bearer token. The README says agent tools are read-only, reindex is off unless enabled, and it lists one-click setup for Claude and Cursor. | Reported |
| 04 | Builds for Windows, macOS and Linux ship on the GitHub releases page with kensaku-cli and a SHA256SUMS file; the README says builds are not yet code-signed and the auto-updater is not wired up. The latest release when we checked was v1.0.11 (Oct 30, 2026). | Reported |
| 05 | The author's own benchmark on Linux reports 1.3 ms average in-memory name search over a 100,000-file folder, about 12 MB of memory for names and metadata, and 184 s for a first index of 101,101 files. The README says there is no telemetry, account or cloud search. | Reported |
reported by the sourcemeasured or run by usnot verified
Primary source: Kensaku repository and README (GitHub). We are the fast, independent index; the vendor’s own page is the source of record for this release.
In the source’s words
“You don't remember the filename. You remember what was in it.”
“Responses are budgeted in size, and file content is returned as data, never as instructions.”
Early verdict
Worth a look if you want agents to search your own files without a cloud index: the README describes a read-only, local MCP server on top of a full-text and OCR index, though builds are still unsigned.
Basis: Based only on the project's README and releases page as read on Nov 1, 2026; we have not run it.
What we have not verified
- We did not install or run Kensaku; every speed, memory and size figure is the author's own measurement from the README.
- We did not test the MCP server with Claude, Cursor or any other agent, or check its read-only and prompt-injection claims.
- The README says Windows gets the most real-world testing and the macOS and Linux builds are less tested; we did not check either.
This brief is Reported: written from the sources below, not from running the release ourselves.
Sources
- Kensaku repository and README (GitHub)primary source · fetched 2026-11-01
- Kensaku releases (GitHub)source · fetched 2026-11-01
Related
Spotted an error? See our corrections policy or email the address on the contact page. Markdown version for agents: /news/kensaku-local-file-search-mcp.md
Want these briefs in your Google Top Stories? Add Agentic Weekly as a preferred source on Google.
The Weekly Verdict · free
5 new-model use cases. Every Monday.
A free email every Monday: the week's few big AI releases, then 5 short use cases for the new models, each with a prompt you can paste.
Free. Double opt-in. One click to unsubscribe.