sgrep
Semantic and hybrid code and conversation search for intent-based queries.
Going through the checklist
What it does
Use sgrep for semantic and hybrid search across code and agent conversations. It understands intent, not just exact strings.
Use when exploring unfamiliar codebases, finding code by concept instead of exact text, or recalling past agent conversations about similar…
You don’t need to name it. Ask for the job in your own words — your AI loads the skill when the request matches.
What it can touch
Includes 159 scripts
Your AI can run these when the skill is used — on your computer in a coding tool (which asks first unless you have told it not to), or in the AI’s own sandbox in Claude and ChatGPT.
How we know
Found in the skill’s folder: bench/jev/coderank_benchmark.py, bench/jev/jev.go, bench/jev/lateon_benchmark.py, bench/jev/main.go….
Uses terminal commands
It shows your AI commands to run. In a coding agent that means your terminal; in a chat app it usually can’t.
How we know
The SKILL.md contains shell code blocks.
Deletes your home folder or the whole drive
Run carelessly — or by an agent working unattended — this can delete work that cannot be recovered.
How we know
scripts/context_sweep.sh line 16: “rm -rf "$home"”. Read 2026-10-04, commit d3776bc.
scripts/context_sweep.sh, line 16 ↗Its code sends data over the internet
A script in it uploads or posts data. That is normal for a skill that talks to a service — check that the address is the service you expect.
How we know
bench/jev/coderank_benchmark.py line 133 (and 1 more in that file): “request = urllib.request.Request(endpoint + "/v1/rerank", data=body, headers={"Content-Type": "application/json"})”. Read 2026-10-04, commit d3776bc.
bench/jev/coderank_benchmark.py, line 133 ↗Asks for administrator rights
Some of its commands use sudo or "run as administrator". Those can change anything on your computer.
How we know
Seen in its code. Read 2026-10-04, commit d3776bc.
7 smaller things worth knowing
Takes instructions from a web page that can change at any time; Downloads a script and runs it in one step, unread; Contains commands that delete or overwrite. Usually legitimate — listed so nothing is hidden.
How we know
Takes instructions from a web page that can change at any time — docs/content/guides/conversation-search.md line 60; Downloads a script and runs it in one step, unread — docs/content/guides/getting-started.md line 20; Contains commands that delete or overwrite — install.sh line 95; Contains commands that delete or overwrite — pkg/conv/store.go line 700; Contains commands that delete or overwrite — pkg/store/fts.go line 27; Contains commands that delete or overwrite — pkg/store/inmem_test.go line 328. Read 2026-10-04, commit d3776bc.
docs/content/guides/conversation-search.md, line 60 ↗docs/content/guides/getting-started.md, line 20 ↗install.sh, line 95 ↗pkg/conv/store.go, line 700 ↗pkg/store/fts.go, line 27 ↗pkg/store/inmem_test.go, line 328 ↗Partly checked
Launch7’s check could not read everything in this skill: 1 file too large to read. Nothing is known about the parts it skipped.
How we know
239 text files read; skipped files are listed by name only. Read 2026-10-04, commit d3776bc.
Recently maintained
Its repository has had changes recently.
How we know
Last push to XiaoConstantine/sgrep: 2026-09-19 (15 days before the scan).
Read automatically from the skill’s own files on GitHub on 4 Oct 2026. Launch7 does not run or manually review skills: this lists what the check saw, and nothing here means “safe”. How we check
Add it to your AI
Partly. Upload it in Claude’s settings — the instructions work; the scripts need the full folder.
Built in your browser from the SKILL.md on GitHub — nothing passes through Launch7.
In Claude, open Customize → Skills, press +, then Create skill → Upload a skill, and choose the .zip.
Make sure code execution is switched on (Settings → Capabilities). Skills need it.
Start a new chat and ask for the job in your own words — Claude picks the skill up when it fits.
This skill ships 159 scripts beside its instructions. The download here is the SKILL.md alone — enough for the instructions, not for the parts that run code. Use the command route, or take the full folder from the repository.
How Claude does this: support.claude.com · checked 28 Sep 2026 · “Supported” means documented by the vendor; Launch7 has not run it end to end