X Research Skill

Agentic research over X/Twitter using the x_search tool. Decompose the research question into targeted searches, iterate to refine signal, and synthesize into a sourced sentiment briefing.

Research Loop

1. Decompose into Queries

Turn the research question into 3–5 targeted queries using X operators:

Core query: Direct keywords or $TICKER cashtag
Expert voices: from:username for known analysts or accounts
Bearish signal: keywords like (overvalued OR bubble OR risk OR concern)
Bullish signal: keywords like (bullish OR upside OR catalyst OR beat)
News/links: add has:links to surface tweets with sources
Noise reduction: -is:reply to focus on original posts; -airdrop -giveaway for crypto topics

2. Execute Searches

Use the x_search tool with command: "search". For each query:

Start with sort: "likes" and limit: 15 to surface highest-signal tweets
Add min_likes: 5 or higher to filter noise for broad topics
Use since: "1d" or "7d" depending on how time-sensitive the topic is
If a query returns too much noise, narrow with more operators or raise min_likes
If too few results, broaden with OR terms or remove restrictive operators

3. Check Key Accounts (Optional)

For well-known analysts, fund managers, or company executives, use command: "profile" to see their recent posts directly.

4. Follow Threads (Optional)

When a high-engagement tweet appears to be a thread starter, use command: "thread" with the tweet ID to get full context.

5. Synthesize

Group findings by theme (bullish, bearish, neutral, news/catalysts):

### [Theme]

[1–2 sentence summary of the theme]

- @username: "[key quote]" — [likes]♥ [Tweet](url)
- @username2: "[another perspective]" — [likes]♥ [Tweet](url)

End with an Overall Sentiment paragraph: predominant tone (bullish/bearish/ mixed/neutral), confidence level, and any notable divergence between retail and institutional voices.

Refinement Heuristics

Problem	Fix
Too much noise	Raise `min_likes`, add `-is:reply`, narrow keywords
Too few results	Broaden with `OR`, remove restrictive operators
Crypto spam	Add `-airdrop -giveaway -whitelist`
Want expert takes only	Use `from:` or `min_likes: 50`
Want substance over hot takes	Add `has:links`

Output Format

Present a structured briefing:

Query Summary: what was searched and time window
Sentiment Themes: grouped findings with sourced quotes and tweet links
Overall Sentiment: tone, confidence, key voices
Caveats: X sentiment is not a reliable predictor; sample bias toward vocal minorities; last-7-days window only

x-research