crux-litsearch skillA
crux-litsearch is agent-read markdown (skill) from mehdiforoozandeh/crux: Set up a literature search over a crux vault with the PI, in two asks and one approval, and write it to `wiki/lit/<slug>/scope.md` — what the search is for, and the 3–8 papers the PI already trusts, which are what the citation crawl walks outward from. Spends its turns getting the seeds right rather than polishing prose, because the seed set defines the subgraph and a beautiful scope over bad seeds gives a bad list. Lints the file with `crux lit lint`, then stops and hands the PI the exact `crux.
Indexed from public GitHub and served as immutable, content-addressed versions. Install it pinned to an exact SHA-256 with the mdr CLI, and every file is verified against the hash recorded here before it reaches your agent. The deterministic audit below grades the latest version, and the same file always earns the same grade.
What the file says
# crux-litsearch — write the scope, then stop A crux wiki's PI curates `raw/` from memory. This skill replaces that with the citation graph's own answer: start from papers the PI already trusts, walk outward, and rank what comes back by how much of the seed set reaches it. You do the conversation and write one file; the engine crawls and ranks. You never run the crawl and never put anything in `raw/` unasked — [crux-wiki](../crux-wiki/SKILL.md)'s rule that the PI alone curates sources holds; they just curate from a computed list, not memory. ## The two asks and the one approval **Ask 1 — what is this search for?** One paragraph in the PI's own words, not a keyword list: it is what they read in a month to remember why these papers. Write what they say, not a tidied version. **Ask 2 — which papers do you already trust?** The ask that matters, and where the turns go. Target **3–8**: fewer and the shared subgraph is whatever one paper happened to cite, more and the seeds stop agreeing on a subject. Take DOIs where the PI has them; where they name a paper instead, resolve it and read the title back before it enters the file — a wrong seed is …
Read the whole file at its exact version.
How to install
mdr add mehdiforoozandeh/crux/crux-litsearch@v1.0mdr add mehdiforoozandeh/crux/crux-litsearch@sha256:621bceeaf4389c84Pin to a label to follow the author's releases, or to a sha256 to freeze the exact bytes forever. Either way the resolved hash is written to mdr.lock, and mdr install reproduces it on any machine.
[](https://markdownregistry.com/a/art_njzt6dx52ok5dmiy)
1 badge views in 30 days
Versions
| version | committed | commit | size | audit | |
|---|---|---|---|---|---|
| v1.0 latest | 2026-09-20 | b301cc8 | 5,045 B | A | view · diff |
| v1.0 | 2026-09-20 | 3b653da | 4,790 B | A | view |
Audit of the latest version
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (5045 bytes)
- pass: No zero-width or bidi control characters
- pass: No instruction hidden inside an HTML comment
- pass: No link to an exfiltration or paste host
- pass: No credential-shaped string
- pass: No instruction to send local credentials anywhere
- pass: No text hidden with inline styles
- pass: No prompt-injection phrasing
- pass: No curl or wget piped into a shell
- pass: No recursive delete of root, home or parent
- pass: No instruction to read or print local credentials
- pass: No base64 blob over 200 characters
- pass: No link to a raw IP address
- pass: No script tag
Source
mehdiforoozandeh/crux · 6 stars · license MIT · pushed 2026-09-21 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_njzt6dx52ok5dmiy GET https://markdownregistry.com/api/v1/resolve?ref=mehdiforoozandeh/crux/crux-litsearch GET https://markdownregistry.com/api/v1/blob/621bceeaf4389c841022d0a76f12dcf0ea5d6caab13964a667632bb3c351d59a
Your agent does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.