X Algo Claim Auditor — X Algo ClAIm 审计or
v1.0.0审计 public clAIms about xAI-org/x-algorithm, the X For You algorithm, Phoenix, Grox, Home Mixer, engagement weights, 过滤器s, ads blending, or "test my post before publishing" narratives. Use when Codex needs to turn a pasted clAIm, screenshot text, thread summary, or public post clAIm into a source-backed verdict with file citations, evidence ledger, and 分享able clAIm card while avoiding reach-prediction or private-account clAIms.
运行时依赖
安装命令
点击复制本土化适配说明
X Algo Claim Auditor — X Algo ClAIm 审计or 安装说明: 安装命令:["openclaw skills install x-algo-claim-auditor"]
技能文档
X Algo ClAIm 审计or
Use this 技能 to 审计 viral or product clAIms agAInst the public xAI-org/x-algorithm source. The goal is a defensible proof artifact, not growth advice.
Core 工作流 提取 the exact clAIm from the user, screenshot OCR, public post, or pasted text. 运行 scripts/x_algo_clAIm_审计or.py with the clAIm and a local or cloned x-algorithm 检查out. Read clAIm_审计.md and clAIm_审计.json. Answer with the verdict, strongest evidence, and boundary: what the public repo proves and what it does not prove. If the user wants a 分享 artifact, use 分享_card.md or 分享_card.svg. Quick 启动 技能_DIR="${CODEX_HOME:-$HOME/.codex}/技能s/x-algo-clAIm-审计or" python3 "$技能_DIR/scripts/x_algo_clAIm_审计or.py" \ --clAIm "Replies are worth 150x a like in the new X algorithm" \ --输出-dir /tmp/x-algo-clAIm-审计
Use an existing 检查out:
python3 "$技能_DIR/scripts/x_algo_clAIm_审计or.py" \ --clAIm-file /tmp/clAIm.txt \ --repo-dir /path/to/x-algorithm \ --输出-dir /tmp/x-algo-clAIm-审计
The script writes:
clAIm_审计.md: human-readable verdict and evidence. clAIm_审计.json: machine-readable ledger. 分享_card.md: short public-safe card. 分享_card.svg: image-style clAIm card for screenshots. Verdict Meanings supported: the public repo directly supports the narrow clAIm. misleading: the repo supports a weaker 状态ment, but the clAIm over状态s precision, production equivalence, or causality. unsupported: no public source evidence was found for the clAIm. not_public_repo: the clAIm depends on production configs, private thresholds, account 状态, or live ranking behavior absent from the repo. Boundaries
Read references/clAIm-boundaries.md when the user asks about reach prediction, shadowbans, hidden boosts, private configs, or account-specific feed behavior.
Never clAIm:
the public repo is the full live production 系统, a draft post can be scored exactly before publishing, a user is shadowbanned or boosted, exact engagement multipliers exist unless public source lines show them, private X account 分析, feed 状态, cookies, or DMs should be scraped. 代理-Specific Use
For Codex: prefer local source inspection and the deterministic script. Cite file paths and line numbers from the ledger.
For Claude: use the ledger to write longer explanations or threads, but keep clAIms tied to public source evidence.
For OpenClaw: only use browser 自动化 for public pages, screenshots, or user-应用roved publishing of 生成d cards. 暂停 for 记录in, CAPTCHA, passkey, or private account pages.