domain-skills(x): add logged-in X web-search workflow

[janzheng]

Adds domain-skills/x/search.md — a workflow for driving the user's logged-in Chrome at x.com/search to read results with the full advanced-search operator set, reaching past the public API's ~7-day recent-search ceiling.

There was no x/twitter domain skill before, so the next harness run on X starts ahead.

Captures:

• URL pattern (?q=...&f=live|top) and the advanced-search operator table (since:/until: for any date window, min_faves:, from:, filter:, lang:, boolean groups).
• Login-wall STOP rule — detect /login redirect and stop, never type credentials.
• Scroll-while-extract guidance for X's virtualized timeline (it recycles off-screen DOM nodes).
• Stable selectors, and the aria-label exact-count trick — visible engagement numbers are truncated ("1.2K") but [role=group][aria-label] carries the exact integers.
• A js() extractor that emits the universal SourceItem shape directly, a full capture snippet, and a traps list.

🤖 Generated with Claude Code

---

Summary by cubic

Add a logged-in X (Twitter) web-search workflow that uses advanced operators and reaches beyond the public API’s ~7‑day window. Includes a robust extractor and guidance for virtualized lists and auth walls.

• New Features
  • Adds domain-skills/x/search.md to drive x.com/search (?q=...&f=live|top) using the logged-in session.
  • Enables full advanced operators in q (since:/until:, min_faves:, from:, filter:, lang:, boolean).
  • Handles auth via a login-wall STOP rule; never types credentials.
  • Provides scroll-while-extract guidance, stable selectors, aria-label exact counts, and a js() extractor + capture snippet.

^{Written for commit d0c35f1. Summary will update on new commits.}

[browser-harness-review]

✅ Skill review passed

Reviewed 1 file(s) — no findings.

[cubic-dev-ai]

1 issue found across 1 file

Prompt for AI agents (unresolved issues)

Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="domain-skills/x/search.md">

<violation number="1" location="domain-skills/x/search.md:80">
P1: Extractor drops media-only tweets by requiring non-empty `tweetText`, breaking `filter:media`/`filter:videos` captures</violation>
</file>

<sub>Reply with feedback, questions, or to request a fix.

Fix all with cubic | Re-trigger cubic</sub>

[cubic-dev-ai]

P1: Extractor drops media-only tweets by requiring non-empty tweetText, breaking filter:media/filter:videos captures

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At domain-skills/x/search.md, line 80:

<comment>Extractor drops media-only tweets by requiring non-empty `tweetText`, breaking `filter:media`/`filter:videos` captures</comment>

<file context>
@@ -0,0 +1,145 @@
+  for (const art of document.querySelectorAll('article[data-testid="tweet"]')) {
+    const textEl = art.querySelector('[data-testid="tweetText"]');
+    const text = textEl ? textEl.innerText.trim() : "";
+    if (!text) continue;
+    const timeEl = art.querySelector('a[href*="/status/"] time');
+    const a = timeEl && timeEl.closest('a[href*="/status/"]');
</file context>