Skip to content

Filter the expired-domain table, and stop claiming sources that never ran - #57

Merged
ThinkingSpade merged 1 commit into
mainfrom
feat/expired-domains-filters-and-reach
Aug 20, 2026
Merged

ThinkingSpade merged 1 commit into
mainfrom
feat/expired-domains-filters-and-reach

Conversation

@ThinkingSpade

Copy link
Copy Markdown
Owner

Both problems came out of the first real run, on the Delio project.

The results were thin, and the panel hid why

Every row said "ranks for N of your keywords" — none said "links to N of your competitors". Link gap contributed nothing, because Delio has 0 competitors saved. That source returned an empty list rather than erroring, so collectCandidates counted it in sourcesUsed and the summary reported "50 checked" as though everything had searched.

That matters more than it sounds: link gap is the source that reaches adjacent domains — a food or nutrition site that links to a vending competitor. Without it the run collapses to SERP rivals over the project's own tracked keywords, which by construction returns more of the same vertical. The output looked like a weak feature; it was actually a feature running on one leg and not saying so.

Now a source declares unavailableReason(context). When it cannot run it is recorded as skipped, not used, and collect is never invoked — so a metered source can never fire as a billed no-op. The panel names what was not searched and points at the fix:

Not searched: link-gap (no competitors saved — link gap is what finds adjacent sites, so add a few on the Competitors tab).

No way to ask "which of these can I actually buy"

Status filters (All / Expired only / Expires soon / This quarter) plus a domain search box. Filtering is client-side over rows already paid for, so changing a filter never re-requests anything. "Filtered to nothing" now reads differently from "found nothing" — those two states were indistinguishable.

Verification

  • pnpm ci:check clean; pnpm test green — 307 files, 3104 tests.
  • New unit tests: filter by status, by substring (case-insensitive), both together, whitespace-tolerant; and the regression that pins the bug — an unavailable source lands in sourcesSkipped, never sourcesUsed, and its collect is not called.
  • e2e/expired-domains.spec.ts still passes: panel mounts idle, quotes its ceiling, zero metered requests before a click.
  • Restore stays compatible — sourcesSkipped defaults to [] so runs recorded before this change still parse.

Still not solved by this PR

Adjacency is only as wide as the candidate sources. With competitors saved, link gap does the work. Reaching genuinely unconnected expired domains in a niche still needs a drop-list feed (DomCop / ExpiredDomains.net) — a second vendor, still out of scope.

🤖 Generated with Claude Code

… ran

Two problems the first real run exposed on a project with no saved competitors.

The link-gap source returned an empty list instead of erroring, so it was
counted in sourcesUsed and the summary reported "50 checked" as though every
source had searched. It had not: link gap is the source that reaches ADJACENT
domains -- a food or nutrition site linking to a vending competitor -- so
without it the run collapses to SERP rivals and returns more of the same
vertical. A source now declares why it cannot run, is recorded as skipped
rather than used, and is never invoked, so a metered source cannot fire as a
no-op. The panel names what was not searched and points at the fix.

And the table had no way to answer "which of these can I actually buy". Status
filters plus a domain search now narrow it client-side, over rows already paid
for, so changing a filter never re-requests anything. Filtering to nothing says
so distinctly from finding nothing -- the two states looked identical before.
@cloudflare-workers-and-pages

cloudflare-workers-and-pages Bot commented Aug 20, 2026 •

Copy link
Copy Markdown

Deploying with  Cloudflare Workers  Cloudflare Workers

The latest updates on your project. Learn more about integrating Git with Workers.

Status Name Latest Commit Updated (UTC)
✅ Deployment successful!
View logs
flyrocketseo 69ec75d Aug 20 2026, 04:35 PM

@ThinkingSpade
ThinkingSpade merged commit 82f6360 into main Aug 20, 2026
3 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant