菜单
Research any topic on X (Twitter) using the last 7 days of public posts. Use when the user asks to research, explore, or find what people are saying about a topic on X/Twitter. Triggers on phrases like "what's trending", "search X for", "what are people saying about", "research on X".
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
Autoresearch loop for governance files. Researches latest X discourse on each governance topic, proposes ONE atomic improvement per file, validates it, keeps or discards. Use when the user asks to improve, update, or evolve the governance framework using latest community insights.
Plan and execute X API retrieval and integration tasks using the official X API docs. Use when the agent needs to determine what X data can be read or managed, map needs like posts, profiles, timelines, bookmarks, likes, follows, lists, spaces, direct messages, trends, usage, or media to the right endpoints, choose auth flows and scopes, and account for fields, expansions, pagination, retention windows, rate limits, or plan limits.
基于 SOC 职业分类
| name | x-research |
| description | Research any topic on X (Twitter) using the last 7 days of public posts. Use when the user asks to research, explore, or find what people are saying about a topic on X/Twitter. Triggers on phrases like "what's trending", "search X for", "what are people saying about", "research on X". |
Search and analyze recent X (Twitter) posts on any topic. Returns engagement-scored results with full tweet data, trending themes, and most-shared links. Uses the X API v2 recent search endpoint (7-day window) with the app-only bearer token from .env.
X_BEARER_TOKEN set in .env (app-only bearer token)Run the research script with the topic as the first argument:
python3 "X-Bookmarks Import/skills/x-research/scripts/x_search.py" "TOPIC" [OPTIONS]
| Flag | Default | Description |
|---|---|---|
--days=N | 7 | Lookback window (max 7, API limit) |
--limit=N | 100 | Max tweets to fetch (paginated at 100/page) |
--no-retweets | off | Exclude retweets |
--lang=XX | any | Filter by language code (en, es, ja, etc.) |
--emit=MODE | full | Output format: full, compact, or json |
# Full research on Claude Code
python3 "X-Bookmarks Import/skills/x-research/scripts/x_search.py" "Claude Code" --no-retweets --lang=en
# Quick scan of AI agents discourse
python3 "X-Bookmarks Import/skills/x-research/scripts/x_search.py" "AI agents" --emit=compact --limit=50
# JSON output for programmatic use
python3 "X-Bookmarks Import/skills/x-research/scripts/x_search.py" "AGENTS.md governance" --emit=json
# Specific user's recent posts
python3 "X-Bookmarks Import/skills/x-research/scripts/x_search.py" "from:anthropaborner" --days=7
full (default)compactjsonPosts are ranked by a composite engagement score:
engagement = likes + (RTs * 2) + (replies * 1.5) + (quotes * 3) + (bookmarks * 2)
velocity = engagement / views * 1000
recency = 1.5 decaying to 0.5 over 7 days
score = (engagement * 0.6 + velocity * 100 * 0.4) * recency
This surfaces high-signal posts (viral + high engagement-to-view ratio) with recency bias.
The topic string supports X search operators:
| Operator | Example | Purpose |
|---|---|---|
from:user | from:AnthropicAI | Posts by specific account |
to:user | to:ClaudeCode | Replies to specific account |
has:media | AI agents has:media | Posts with images/video |
has:links | Claude Code has:links | Posts with URLs |
url:"domain" | url:"github.com" | Posts linking to specific domain |
-is:retweet | auto via --no-retweets | Exclude retweets |
is:reply / -is:reply | Include/exclude replies | |
"exact phrase" | "Claude Code skills" | Exact match |
(A OR B) | (Claude OR Codex) skills | Boolean OR |
The --emit=json output can be piped to other tools or saved for analysis:
python3 .../x_search.py "topic" --emit=json > /tmp/x_research.json
The x-api-data-access skill provides the auth/endpoint reference if you need to extend this.