zero-click-risk-scorer

逐关键词抓取实时 SERP，检测 AI 概览、精选摘要、知识面板等吸收点击的特征，按权重累加得到 0–100 零点击风险分（含首位自然结果被下压的惩罚），分类后计算调整后机会值并按机会降序重排，输出符合 schema 的 JSON；需 SerpApi 密钥。

未知作者

热度 0.00

最近更新 2026-10-18 21:44:31

分类：营销依赖：代码执行 / 第三方 API / 读写文件应用场景：市场分析 / 研究 / 报告撰写 / 数据处理

安装技能

复制下面的话术发给你的 Agent，它会通过 secure-skills 检索并安装该技能。

执行命令安装此 skill： npx secure-skills-cli add seoskills.sh/zero-click-risk-scorer

Metadata

name: zero-click-risk-scorer
description: Analyzes each keyword's live SERP for click-absorbing features and scores its zero-click risk, then reprioritizes the keyword list toward terms that still send organic clicks. Use when the user wants to filter keywords worth targeting, spot high-volume zero-click traps, or understand how AI Overviews and featured snippets erode click potential.
metadata: {"title":"Zero-Click Risk Scorer","category":"keyword-research"}

Zero-Click Risk Scorer

AGENT ROLE: Autonomous SERP-economics agent. Score each keyword's likelihood of resolving without a click and re-rank the list by remaining click opportunity. Emit the JSON in references/output.schema.json.

OBJECTIVE

Assign each keyword a 0–100 zero-click-risk score from the click-absorbing features present on its SERP, and compute an adjusted opportunity so high-volume keywords that no longer earn clicks are deprioritized.

INPUTS

keywords (REQUIRED): array of { term, volume? } (volume optional; if absent, opportunity uses volume=1).
weights (OPTIONAL): override references/feature_weights.json.
location/hl/device (OPTIONAL): default "United States"/"en"/"mobile" (mobile shows more zero-click features).

AUTHENTICATION (SERP API)

REQUIRE env SERP_API_KEY. IF unset THEN STOP error.code="AUTH_MISSING_API_KEY".
Endpoint (SerpApi shape): GET https://serpapi.com/search.json?engine=google&q={term}&device={device}&api_key={key}. AI Overviews may need the follow-up engine=google_ai_overview expansion.

EXPECTED TOOL CALLS

Run scripts/zero_click.py --keywords keywords.json.
Per keyword: fetch SERP; detect which weighted features are present and whether the top organic result sits above or below them.

PROCEDURE (deterministic, per keyword)

STEP 1 — FETCH SERP; detect features from feature_weights.json (ai_overview, featured_snippet, knowledge_panel, people_also_ask, instant_answer/answer_box, inline_videos, local_pack, shopping). STEP 2 — SCORE: risk = min(100, sum(weight for each present feature)). Cap at 100. STEP 3 — POSITION MODIFIER: IF the first organic result is pushed below ≥2 blocks (features stacked above) THEN add the configured push_down_penalty (still capped at 100). STEP 4 — CLASSIFY low (<25) | moderate (25–55) | high (>55). STEP 5 — ADJUSTED OPPORTUNITY: adjusted = round(volume * (1 − risk/100)) — the estimated click-earning potential that survives the SERP. STEP 6 — EMIT keywords sorted by adjusted desc (best real opportunities first); include the present-feature list per keyword as evidence.

RATE LIMITS & ERROR HANDLING

SERP API bills per search. 429/quota → backoff 2^attempt (max 5) then STOP error.code="RATE_LIMITED" with partial. Never drop keywords silently.
5xx/timeout retry ≤3 then mark keyword status="serp_error" (excluded from ranking, listed in skipped).
Concurrency ≤ 3.

MISSING / INSUFFICIENT DATA

IF a feature block is present but its contents are gated by the SERP tier THEN still count its presence for risk (presence is the signal).
IF volume absent THEN report risk but mark adjusted as null (cannot rank by opportunity without volume) and sort those by risk asc.
Never invent volume.

OUTPUT

One JSON object per references/output.schema.json. No prose.

FILES

scripts/zero_click.py — SERP fetch, feature detection, risk + adjusted-opportunity scoring.
references/feature_weights.json — per-feature zero-click weights.
references/output.schema.json — output contract.