x-search

---

name: x-search description: Search X (Twitter) posts via Grok API and ingest results into the knowledge graph. disable-model-invocation: false argument-hint: [search query] [--project project_name] [--days 7] [--handles user1,user2] [--web-search] allowed-tools: Bash, Read, Write, Glob

X (Twitter) Search Ingestion

Search X (Twitter) posts in real-time using xAI's Grok API, save results as structured Markdown, then ingest into the knowledge graph with entity extraction.

Usage

/x-search "AI agent frameworks"
/x-search "product launch" --days 30 --handles elonmusk,openai
/x-search "knowledge graphs" --web-search --project project_a
/x-search "生成AI トレンド" --days 3

Prerequisites

• pip install openai
• XAI_API_KEY environment variable set (get key at https://console.x.ai/)

---

Workflow (Interactive Only)

Note: No --auto mode is provided due to per-search API costs ($0.015-$0.025/call). All searches require user review before ingestion.

1. Execute X Search

Parse $ARGUMENTS for the query and optional flags. Run:

python ${CLAUDE_SKILL_DIR}/../../../scripts/x_search.py "<query>" --days <N> [--handles h1,h2] [--exclude-handles h1,h2] [--web-search] $1 $2

Capture the output path (last line of stdout).

2. Review Results

Read the generated markdown file using the Read tool. Present a summary to the user:

• Key themes and topics found
• Notable accounts/posts mentioned
• Date range covered
• Number of citations

Ask the user if they want to proceed with ingestion into the knowledge graph.

3. Ingest Markdown into Graph

python ${CLAUDE_SKILL_DIR}/../../../scripts/auto_ingest.py upsert "<md_file_path>" $1 $2

Note the chunk count from the output.

4. Extract Entities (Claude Code = Claude)

Read the generated markdown file using the Read tool. Analyze the X search results and extract entities and relationships as JSON.

Entity types:

• PERSON, ORGANIZATION, TECHNOLOGY, REQUIREMENT, SCHEDULE, BUDGET, RISK
• PROPOSAL_PATTERN, EVALUATION_CRITERIA, DELIVERABLE, SECURITY, DOMAIN, CONCEPT

X-search-specific extraction guidance:

• X posts often reference people (@handles) — extract as PERSON entities
• Company and product mentions — extract as ORGANIZATION or TECHNOLOGY
• Trending topics and hashtags — extract as CONCEPT if substantive
• Avoid extracting generic social media terms (retweet, like, thread)

5. Detect Entity Conflicts

Before saving, check for conflicts with existing entities:

python -c "
import sys, json
sys.path.insert(0, '.')
from neo4j import GraphDatabase
from config import get_config
from scripts.save_entities import query_existing_entities
cfg = get_config()
driver = GraphDatabase.driver(cfg.neo4j_uri, auth=cfg.neo4j_auth)
names = [COMMA_SEPARATED_ENTITY_NAMES]
result = query_existing_entities(driver, names)
driver.close()
for name, info in result.items():
    print(json.dumps({'name': name, 'type': info['type'], 'description': info['description'], 'sources': info['sources']}, ensure_ascii=False))
"

For each entity that already exists:

• No conflict (additive or equivalent): Proceed. Use the more comprehensive description.
• Conflict detected (contradicting facts): Present both to the user and let them decide.

6. Save Entities to Graph

cat <<'ENTITIES_JSON' | python ${CLAUDE_SKILL_DIR}/../../../scripts/save_entities.py --source-path "<md_file_path>" $1 $2
{"entities": [...], "relationships": [...]}
ENTITIES_JSON

Note: Community detection and relationship discovery run automatically via the post-entity-save hook.

7. Report Results

Report: search query, date range, markdown output path, chunk count, entity count, relationship count, and any conflicts resolved.

---

Comparison with Other Skills

Feature	/x-search	/youtube-markitdown	/pdf-markitdown	/add-knowledge
Input	Search query	YouTube URL	PDF file	Free text
Source	X posts (real-time)	YouTube video	Document file	User input
API cost	~$0.02/search	Free	Free	Free
Automated mode	No	Yes (--auto)	Yes (--auto)	No
Best for	Trends, discussions	Video content	Structured docs	Quick notes

Notes

• Requires: pip install openai and XAI_API_KEY environment variable
• Default model: grok-4-1-fast-non-reasoning (cheapest: $0.20/M input, $0.50/M output)
• Tool cost: $5 per 1,000 calls (~3-5 calls per search)
• --web-search adds web context alongside X results (optional)
• --handles and --exclude-handles cannot be used together (API constraint)
• Output saved to docs/x_search_{query}_{date}.md