docs(domain-skills): add X (Twitter) articles + tweets reading recipe

[optemism]

What

Adds domain-skills/x/articles.md — a field-tested recipe for reading X Articles (long-form posts) and regular tweets.

Key findings captured

• X Articles share the tweet URL shape (x.com/{handle}/status/{id}) — there is no /article/ route; document.title tells you which you got.
• The "See what's happening" login modal blocks clicks, not DOM reads — the article body is fully hydrated underneath and document.body.innerText reads straight through it. (Observed with a logged-in profile; fully logged-out is flagged as unverified.)
• [data-testid="tweetText"] is empty for Articles — it works for regular tweets/threads only; Articles need the innerText path.
• Tab-drift trap: navigate + extract must happen in ONE browser-harness -c invocation, or a follow-up call can attach to a stale/different tab. ensure_real_tab() referenced as the canonical remedy.
• Hydration is lazy — wait_for_load() alone returns a short/empty body; needs wait(3–4).
• Body-slicing helper anchored on the post-timestamp regex, with noted failure modes.

Follows the SKILL.md domain-skill conventions: no pixel coordinates, no run narration, no secrets; explicit "does not work / untested" section.

🤖 Generated with Claude Code

---

Summary by cubic

Adds domain-skills/x/articles.md, a concise recipe for reading X (Twitter) Articles and regular tweets via DOM text extraction. Covers URL shape, selectors, timing, and tab attachment to make extraction reliable even with the login modal.

• New Features
  • Articles share the tweet URL x.com/{handle}/status/{id}; use document.title to distinguish.
  • For Articles, read document.body.innerText; the login modal blocks clicks but not reads.
  • For tweets/threads, use [data-testid="tweetText"] (empty for Articles).
  • Avoid tab drift by navigating and extracting in one invocation; wait 3–4s for hydration and use a timestamp-anchored slice to isolate the body.

^{Written for commit 49a9922. 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/articles.md">

<violation number="1" location="domain-skills/x/articles.md:64">
P2: `x_article_body` can silently return a near-empty string when no newline follows the headline. When `full_text.find('\n', i)` returns `-1` (no newline after the headline), `start` becomes `-1`. Python slicing `full_text[-1:]` then evaluates to just the last character of the page, the regex fails, and the function returns a single character instead of the article body. Add a guard so `start` falls back to `i + len(headline)` when no newline is found.</violation>
</file>

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

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

[cubic-dev-ai]

P2: x_article_body can silently return a near-empty string when no newline follows the headline. When full_text.find('\n', i) returns -1 (no newline after the headline), start becomes -1. Python slicing full_text[-1:] then evaluates to just the last character of the page, the regex fails, and the function returns a single character instead of the article body. Add a guard so start falls back to i + len(headline) when no newline is found.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At domain-skills/x/articles.md, line 64:

<comment>`x_article_body` can silently return a near-empty string when no newline follows the headline. When `full_text.find('\n', i)` returns `-1` (no newline after the headline), `start` becomes `-1`. Python slicing `full_text[-1:]` then evaluates to just the last character of the page, the regex fails, and the function returns a single character instead of the article body. Add a guard so `start` falls back to `i + len(headline)` when no newline is found.</comment>

<file context>
@@ -0,0 +1,131 @@
+def x_article_body(full_text, headline):
+    # start just after the headline's first occurrence in the content area
+    i = full_text.find(headline)
+    start = full_text.find('\n', i) if i != -1 else 0
+    # end at the post timestamp ("H:MM AM/PM · Mon DD, YYYY") — search AFTER
+    # start, or a timestamp-shaped string earlier in the page truncates the body
</file context>