Structure drifts
Fields, naming, and categories diverge as more people contribute.
Open-source framework for maintained knowledge spaces
Build community knowledge that stays useful.
Grove turns community submissions into structured, searchable, and maintainable knowledge spaces—using portable files, Git-based review, and static publishing.
Astro renderer availableMIT licensedActive development
●data/records/immich.yml
kind: project
name: Immich
category: media
repoUrl:
https://github.com/immich-app/immich●Grove
- 01 Validate
- 02 Enrich
- 03 Review
- 04 Generate
●Generated output
/projects/immichrecords.index.jsonsitemap.xmlllms-full.txt
One reviewed record, ready for people, search engines, and AI systems.
100Performance
100Accessibility
100Best Practices
100SEO
Measured on this landing page’s production build. Custom content and third-party integrations can affect individual scores.
$
pnpm dlx @grove-dev/cli@latest new my-spaceThe maintenance problem
Lists are easy to start and difficult to maintain.
A README or small static site works until the collection becomes shared infrastructure. Growth turns publishing into recurring maintenance work.
Metadata goes stale
Archived projects and old releases remain visible without a repeatable refresh.
Review becomes cleanup
Maintainers normalize every submission before they can judge its value.
Discovery becomes software
Search, filters, routes, metadata, and sitemaps become another app to maintain.
Machines receive fragments
A README is readable, but it is not a predictable record set or generated context.
Maintenance lives in memory
Quality depends on someone remembering what needs review and why.
How Grove works
One lifecycle, with maintainers in control.
Grove automates repeatable checks and generation. It does not automate editorial judgment.
grove — maintained knowledge lifecycle
✓ A community suggestion
Collect
- Grove does
- Import a Markdown list or add a YAML record.
- Maintainer controls
- Maintainers define what belongs.
Output
data/records/<slug>.yml
$ pnpm exec grove import <awesome-list-url>✓ An unstructured candidate
Structure
- Grove does
- Map fields to the selected blueprint and taxonomy.
- Maintainer controls
- Maintainers own categories and wording.
Output
A typed project, resource, or entity
$ kind: project
category: self-hosted✓ Records + taxonomy files
Validate
- Grove does
- Check schemas, slugs, decisions, health, and references.
- Maintainer controls
- Warnings can become strict CI failures.
Output
Actionable validation results
$ pnpm exec grove validate --strict✓ A project with a repository URL
Enrich
- Grove does
- Fetch activity, releases, license, topics, and archive state.
- Maintainer controls
- Synced signals remain reviewable data.
Output
Current repository metadata
$ pnpm exec grove sync github✓ A proposed record change
Review
- Grove does
- Run checks in Actions and review the pull request.
- Maintainer controls
- A human makes the editorial decision.
Output
Accepted, revised, or rejected change
$ git diff data/records/immich.yml✓ Reviewed records
Publish
- Grove does
- Build JSON, sitemap, AI-readable files, and Astro pages.
- Maintainer controls
- Choose the host, routes, and presentation.
Output
Static, cacheable output
$ pnpm exec grove generate && pnpm build✓ A growing, aging collection
Maintain
- Grove does
- Report stale, inactive, archived, and incomplete records.
- Maintainer controls
- Humans decide to keep, hide, or remove.
Output
A focused maintenance report
$ pnpm exec grove cleanup stale --reportOne source, multiple outputs
Maintain one record. Serve people and machines.
Grove generates a website for people, crawlable pages and a sitemap for search engines, structured JSON for other tools, and predictable context files for AI systems.
It does not promise rankings or AI visibility. It produces stable, inspectable formats from the same reviewed source.
- static HTML
- sitemap.xml
- llms.txt
- llms-full.txt
record.yml
name: Immich
kind: project
category: media
Website
Static pages
Search
sitemap.xml
AI
llms.txt
Tools
records.json
Production reference
Grove grew out of maintaining Open Apps.
Open Apps is a live directory with file-backed records, repository refresh, search, filters, detail pages, a sitemap, and AI-readable outputs. Grove turns that operating model into a reusable framework.
The live repository still carries local versions of this logic. Migration to the published Grove packages is the next reference-space milestone.
open-apps.dev.mn/apps
Open Apps
Explore real open-source apps
SearchFilter
ImmichMedia
TypeScriptactiveJoplinProductivity
TypeScriptmatureLinkwardenKnowledge
Next.jsactivetaxonomysitemapllms.txt
What Grove can power
For collections that need to stay useful.
Grove is not only a directory template. Its record model supports three typed blueprints; the polished default frontend is available for project directories today.
Project directories
Developer tools, open-source apps, packages, services, or internal systems.
Resource hubs
Guides, comparisons, explainers, learning collections, and curated links.
Ecosystem maps
Organizations, products, communities, schools, and local ecosystem actors.
Maintained awesome lists
Import a Markdown collection, then add structure, review, and static discovery.
Architecture and status
Headless lifecycle. Replaceable frontend.
Core owns the source model and generated data. UI owns framework-neutral primitives. An adapter renders the site. Your repository keeps the content, brand, and editorial rules.
Supported today
Astro
V1 supported renderer
Planned adapters
SvelteKit
Skeleton · planned V1.1
Next.js
Skeleton · planned V1.2
The CLI refuses --framework {svelte|nextjs}until the adapter ships components, layouts, and a working template.
Package boundaries
@grove-dev/core
Schema + lifecycle
@grove-dev/ui
Headless primitives
@grove-dev/cli
Scaffold + commands
Start with the Astro implementation.
Inspect the example, read the architecture, and build a space from plain files. No CMS or runtime database required.
$pnpm dlx @grove-dev/cli@latest new my-space