메뉴
Aggregate, score, and prioritize social mentions across X/Twitter. Outputs a ranked inbox with engagement scores and draft context for efficient reply batching.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Encode EVM function calls and constructor args from signature + values. Zero deps.
Find Uniswap V2/V3 liquidity pools for any token on Base or Ethereum — reserves, pricing, TVL
Inspect Uniswap Permit2 sub-approvals for any wallet on Base or Ethereum
Resolve EVM function selectors (4-byte) and event topic hashes to human-readable signatures. Queries openchain.xyz + 4byte.directory with a built-in fallback DB of 50+ common signatures. Batch mode, bytecode scanning, stdin pipe. Zero dependencies.
Resolve Basenames (.base.eth) to addresses and back on Base
Turn raw EVM revert data into a readable error. Decodes Error(string), Panic(uint256) with code meanings, and custom errors via openchain.xyz selector lookup. Zero dependencies. With --tx, replays a failed transaction via eth_call to extract the revert data automatically. Use when an agent's tx reverted and the only signal is an opaque 0x… blob.
| name | social-inbox |
| description | Aggregate, score, and prioritize social mentions across X/Twitter. Outputs a ranked inbox with engagement scores and draft context for efficient reply batching. |
Fetches X/Twitter mentions and brand searches, deduplicates against already-replied tweets, scores by engagement potential, and outputs a prioritized inbox.
# Full scan: mentions + brand search
node ~/Github/axiom-public/agent-skills/skills/social-inbox/social-inbox.mjs
# Mentions only
node ~/Github/axiom-public/agent-skills/skills/social-inbox/social-inbox.mjs --mentions-only
# Output as JSON (for piping)
node ~/Github/axiom-public/agent-skills/skills/social-inbox/social-inbox.mjs --json
# Custom search terms
node ~/Github/axiom-public/agent-skills/skills/social-inbox/social-inbox.mjs --search "AppFactory,axiom ventures"
Saves to ~/clawd/data/social-inbox.json with structure:
{
"lastScan": "2026-02-08T...",
"items": [
{
"id": "tweet_id",
"author": "@handle",
"text": "...",
"url": "https://x.com/...",
"score": 8.5,
"scoreBreakdown": { "relevance": 3, "question": 2, "directMention": 2, "recency": 1.5 },
"status": "pending",
"source": "mentions"
}
]
}
| Signal | Points |
|---|---|
| Direct @mention | +2 |
| Question mark (asking us something) | +2 |
| Keywords: fund, invest, agent, build, LP, token | +1 each (max 3) |
| Recency (<1h: +2, <6h: +1.5, <24h: +1) | up to +2 |
| Not a bot/spam pattern | +1 (default, -1 if detected) |
Items scoring <3 are filtered out. Already-replied tweets are excluded.
Add to OpenClaw cron for periodic scanning:
{
"kind": "agentTurn",
"message": "Run social inbox scan: node ~/Github/axiom-public/agent-skills/skills/social-inbox/social-inbox.mjs\nReview the top 3 items and draft replies if valuable."
}
twitter-api.py (already installed)twitter-replied.json dedup log (already exists)