Zigistry Scraper

Scrape Zigistry — the community-maintained index of packages and programs written in the Zig programming language. Search or browse by topic, fetch full details for specific repositories (dependencies, releases, dependents), or pull curated trending lists. Pure HTTP against Zigistry's public JSON API — no auth, no proxy, no cookies.

What this actor does

• Three modes: search (full-text + topic browse), byRepo (exact repo detail lookup), trending (curated lists)
• Covers both catalogs: reusable packages (libraries) and programs (applications)
• Filters: minimum stars, source host (GitHub/Codeberg), primary language, license (SPDX identifier), exclude archived repos
• Sorting: stars, dependents, recently updated, newly added, name, forks, issues, minimum Zig version
• Empty fields are omitted — you only ever see fields with real data

Output per repository

• id — canonical repo id (e.g. gh/hejsil/zig-clap)
• provider — github or codeberg
• ownerName, repoName, fullName
• description, primaryLanguage, license
• stargazerCount, forkCount, watchersCount, issuesCount, dependentsCount
• isArchived, isDisabled, isFork
• defaultBranchName, createdAt, pushedAt
• minimumZigVersion
• avatarUrl, repoUrl, sourceUrl (the repo's GitHub/Codeberg page)
• mode=byRepo only: latestVersion (newest release, always populated even when version is pinned), resolvedVersion (present only when version was pinned to something other than the latest), publishedAt, readmeUrl, isPrerelease, directoryFiles[], dependencies[] ({name, url, hash, lazy, path}), releases[], releaseCount, dependents[] (capped at 25)
• recordType: "package" / "program", scrapedAt
• mode=trending only: trendingCategory — which curated list (recently_updated / most_used / latest / web / gui / games) this record came from. When category is left blank (all categories), a repo that appears in more than one curated list is emitted once per list, each tagged with its own trendingCategory — use this field to tell those apart.

Input

Field	Type	Default	Description
mode	string	search	search / byRepo / trending
packageType	string	packages	packages / programs
searchQuery	string	–	Free-text query (mode=search)
topic	string	–	Filter by topic, e.g. web, gui, game
sortBy	string	intelligent	stars / dependents / recently_updated / newly_added / name / forks / issues / zig_version (name and zig_version are sorted client-side — see note below)
sortDir	string	desc	asc / desc
repos	array	–	Exact owner/repo (or gh/owner/repo, cb/owner/repo, full URL) list (mode=byRepo)
version	string	–	Exact release tag for repos lookups
category	string	(all)	Trending category (mode=trending)
provider	string	–	Filter by source host
minStars	int	–	Minimum star count
language	string	–	Filter by primary language (exact match, case-insensitive), e.g. Zig, C, Rust
license	string	–	Filter by SPDX license identifier (exact match, case-insensitive), e.g. MIT, Apache-2.0, GPL-3.0
excludeArchived	bool	false	Drop archived repos
maxItems	int	50	Hard cap (1–2000)

Example: popular actively-maintained web packages

{
  "mode": "search",
  "topic": "web",
  "sortBy": "stars",
  "excludeArchived": true,
  "minStars": 10,
  "maxItems": 30
}

Example: full detail on specific repos

{
  "mode": "byRepo",
  "repos": ["hejsil/zig-clap", "floooh/sokol-zig", "cb/foreverzer0/klack"]
}

Use cases

• Discover actively-maintained Zig libraries for a given domain (web, GUI, games)
• Audit a package's dependency tree and dependents before adopting it
• Track newly-added or trending Zig projects
• Build a curated "awesome-zig" style directory automatically

FAQ

Does this require a Zigistry account or API key? No — Zigistry's package index is fully public.

What's the difference between packages and programs? Packages are reusable libraries meant to be depended on; programs are standalone applications. Zigistry maintains separate catalogs for each — set packageType accordingly.

What does sourceUrl point to? The repo's own GitHub or Codeberg page (repoUrl, duplicated as sourceUrl). Zigistry's own per-repo page is a client-rendered route with no server-side page to fetch, so it isn't used as a URL field here.

Why does sortBy: name / sortBy: zig_version take longer than other sort options? Zigistry's own API silently ignores those two sort keys server-side (confirmed empirically — asc/desc return different but equally unsorted order). To still deliver a correctly-sorted result, this actor fetches every matching record across all pages, sorts client-side, and then emits the top maxItems. Every other sort key (stars, dependents, recently_updated, newly_added, forks, issues, intelligent) is honored natively by Zigistry and streams page-by-page as usual.

Why is resolvedVersion sometimes present and latestVersion sometimes absent? When you pin version to an older release, Zigistry blanks its own latest_version field for that response — this actor backfills latestVersion from the repo's release list so it's always the true newest release. resolvedVersion is only added when the pinned version differs from the latest, so you can tell at a glance whether dependencies/readmeUrl/etc. describe an older release.