We’re building Ever Works in public. Some things may be incomplete, missing, or broken while we continue improving the platform.We’re building Ever Works in public — expect a few rough edges.
Use Case
An "awesome" list is only valuable while it is current — and keeping one current is thankless work. Ever Works builds an awesome-style list in any topic, then maintains it: re-checking links, adding new entries, and pruning dead ones, on a schedule, with everything committed to your Git so the community can contribute too.
Build an awesome-* list for any niche or technology.
Dead links detected and pruned automatically.
The platform finds and adds what is new.
Git-backed, so contributions flow through pull requests.
Pair AI maintenance with community contributions: incoming pull requests are triaged and merged, while the runtime keeps the catalog fresh between contributions.
Two ways in — bring the list you already have, or start one that does not exist yet — and the same upkeep loop afterwards. Every screen and button below is named the way the product names it.
To bring a list under management, open New Work → Import Existing Work, paste the repository under Paste URL (or pick it out of My Repositories) and press Analyze Repository. A curated list comes back badged Awesome List Detected, with an estimated item and category count, before you have committed to anything. Starting from nothing instead? Click + New in the sidebar, describe the list — “awesome list of self-hostable open-source SaaS alternatives, categorized and docker-ready” — and pick the Awesome Repo chip.
Import & Research mode does not copy the source. It treats every link as a seed, visits each item’s own page to write an original description, then goes looking for what the list is missing — alternatives, newer projects, whole categories nobody added. The Expansion target sits on the recommended 2.5x, which puts the source at no more than 40% of the finished collection, and Rewrite descriptions, Expand categories & tags and Parse PRs & issues are toggles you set before pressing Start Import & Research. What lands is a regenerated README: a table of contents with a count beside every category, and each entry rendered as a link, an original one-line description, its tags, and a Read more pointing at that item’s own details page.
Switch community pull requests on and, every hour, the platform sweeps the open PRs on the repository, reads each diff with your list’s existing categories in hand, extracts the structured item, commits it to the data repository with a message linking back to the PR, and comments on the thread listing exactly what it added — closing the PR for you if you left auto-close on. The Pull requests tab shows everything open across the data, main and website repositories, and Request agent review puts a written review on the ones you want a second opinion on before merging.
An Awesome README import arms a weekly sync for you with Create Pull Requests already on, so every re-run arrives as a reviewable pull request instead of a surprise commit; pass sync: false at import time if you would rather it did not. On the Work’s Schedule you can pick any of the seven cadences from Every hour to Every month, set a Failure limit that auto-pauses the automation after that many consecutive failures — three by default, after a 15-minute retry — and press Run now whenever you like. Source Health Checks run on their own Check frequency alongside it.
A list you can maintain is a list you can read as files. The data repository is the source of truth: categories.yml, tags.yml and collections.yml at the root, then a folder per item holding a small YAML file — name, description, source_url, category, tags, featured — and an optional long-form markdown body. The README repository is the artefact people actually read on GitHub: README.md, a licence, and a details page per entry. It is published to your Git provider and never deployed anywhere, which is exactly right for a list that lives as a repository. A website repository is provisioned alongside them, so the day you want a browsable, searchable site instead of a scroll, that is a deploy rather than a rebuild.
Every Work kind gets a different set of tabs and settings. This one keeps the curation machinery a list actually needs, and leaves off the one directory affordance that never made sense for it.
A plain dead-link checker catches half of what rots. Every entry gets a reachability verdict — Healthy, Broken link, Could not verify — and a separate accuracy verdict, so a URL that still resolves but has quietly become the vendor’s homepage shows up as Source too generic rather than passing. Where the checker found something better, Apply suggestion swaps it in; Re-check source runs one item on demand.
Contributors keep the workflow they already know — open a PR against the list — and stop being blocked on you finding an evening. The hourly processor does the transcription: parse the diff, categorize the item, commit it, comment with what was added. You review the result, not the formatting.
Categories and tags are editable in the dashboard as well as in the repo, Collections give you hand-picked groupings with an icon and a priority for the entries that deserve the top of the page, and bulk item import and export is there for the day you need to move a few hundred rows at once. Head-to-head comparison pages are the one directory feature this kind deliberately leaves off.
An Awesome Repo Work starts from the Classic Next.js base, the same one directories use. In the public blueprint catalog the Awesome List entry is still marked as a placeholder and is filtered out of the picker, so what you get today is the chip, the import-and-research path and the whole upkeep loop — not yet a bespoke awesome-list theme.