GEO Optimizer helps brands earn citations and recommendations in generative AI search answers through research-driven content placement.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install auto-geo
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install auto-geo using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is GEO Optimizer?
GEO Optimizer is an AI search optimization workflow for improving how products and brands appear in answers from Doubao, Kimi, DeepSeek, Yuanbao, Metaso, Perplexity, and similar generative engines. Given a topic or product name, it researches realistic user questions, identifies cited sources, evaluates publisher authority and accessibility, and recommends the highest-value publishing outlets.
As part of Openclaw Skills, it turns the research into structured, quotable, FAQ-ready Markdown for each selected platform. The workflow combines SEO information gathering, source-backed writing, compliance review, platform adaptation, publishing instructions, and post-publication citation tracking.
GEO Optimizer Use Cases
- Improve brand visibility in AI search and generative engine results.
- Discover which websites and content formats AI engines currently cite for a topic or product.
- Prioritize publishing platforms using citation frequency, authority, topical relevance, and publishability.
- Create platform-specific content for long-form articles, educational posts, conversational short-form content, and technical tutorials.
- Build SEO briefs covering keywords, long-tail queries, audiences, differentiators, proof points, trust signals, and calls to action.
- Produce citation-friendly answers with clear conclusions, facts, definitions, source notes, and FAQs.
- Diagnose existing GEO performance without generating new content by running only the research and source-analysis stages.
- Recheck AI citation coverage two to four weeks after publication and iterate the content strategy.
How GEO Optimizer Works
- Interpret the request: If the user provides a topic or product and asks for GEO, AI search optimization, or brand recommendations, the skill starts the full five-stage workflow. If the user only asks what GEO means, it provides an explanation; if they request diagnosis only, it runs stages one and two.
- Research real user questions: Generate three to six queries from a user perspective, such as product evaluations, competitor comparisons, category recommendations, and official-entry searches.
- Collect cited sources: Search multiple AI engines and general web results when necessary, recording domains, URLs, titles, content types, citation frequency, and the cited context. If an engine cannot be accessed directly, report the public-web substitute used.
- Analyze source outlets: Evaluate each source for official website information, creator-center access, publishing availability, submission model, domain authority, platform search weight, and topical fit.
- Rank publishing territories: Produce a prioritized outlet list that balances citation frequency, authority, relevance, and whether the user can realistically publish there.
- Collect missing SEO inputs: Ask for all missing information in one batch, including core and long-tail keywords, target audience, decision journey, differentiators, evidence, credentials, sales data, case studies, CTA, and compliance restrictions.
- Generate platform-specific content: Create a Markdown document for every selected outlet using the outlet's style while preserving factual accuracy. Typical structure includes a keyword-focused title, TL;DR conclusion, H2/H3 sections, dated source-backed facts, and FAQs.
- Review compliance and publishing requirements: Check advertising-law restricted terms, platform rules, medical and financial claims, disclosure requirements, formatting, tags, timing, and publishing notes.
- Deliver and track: Return source analysis, outlet priorities, content assets, publishing guidance, direct-publishing options, and a recommended follow-up research cycle.
GEO Optimizer Setup
- Install or enable the SkillHub CLI using the official instructions at https://skillhub.cn/install/skillhub.md. The skill documentation specifies installing only the CLI during initial setup.
- Load
fore-vip-geo-optimizerin an Openclaw-compatible agent environment that supports web search or equivalent internet access. - Provide a topic or product name as the initial input. For stronger outputs, also provide available product materials, URLs, proof points, target audiences, competitors, and compliance constraints.
- Enable
web-accesswhen browser-level rendering, authenticated sessions, or direct web verification are required. Standard WebSearch or an internet-search tool is sufficient for ordinary source discovery. - For WeChat Official Account draft delivery, connect the
wechat-oa-draft-pushintegration after content generation. The GEO skill can provide publishing instructions even when direct publishing is unavailable. - If the user requests paid execution steps, install and launch the separate
fore-vip-autoskill through the SkillHub CLI, using the team package@org-2lxcf7z6/fore-vip-auto. GEO delegates that workflow rather than implementing payment, checkout, write-back, or refund logic itself.
# Use the official SkillHub CLI installation instructions
# https://skillhub.cn/install/skillhub.md
# The optional execution skill is installed through the SkillHub CLI:
# @org-2lxcf7z6/fore-vip-auto
GEO Optimizer Data Schema & Taxonomy
The skill organizes each GEO project around a topic or product and produces a research-to-publication report.
| Data area | Stored or generated information |
|---|---|
| Project input | Topic or product name, brand context, URLs, competitors, audience, decision journey, and business goals |
| Query set | Three to six realistic user questions covering evaluation, comparison, recommendations, and official-entry intent |
| Citation source list | Domain, URL, title, cited excerpt or context, content type, appearance frequency, and initial authority assessment |
| Platform analysis | Main site, creator-center URL, publishing model, publishability status, platform weight, topical relevance, and verification notes |
| Priority matrix | Ranked outlets with priority level, selection rationale, publishing feasibility, and recommended content angle |
| SEO brief | Core keywords, long-tail terms, competitor terms, audience, differentiators, evidence, trust signals, CTA, and restricted claims |
| Content assets | Platform-adapted Markdown documents with title, TL;DR, headings, source-dated facts, FAQs, disclosures, and publishing recommendations |
| Publishing metadata | Channel, column or category, tags or hashtags, suggested first-publish timing, review requirements, and manual operating steps |
| Tracking data | Follow-up query set, publication date, citation results after two to four weeks, missed-platform causes, and iteration recommendations |
Reference materials include @references/platforms.md for platform data, @references/seo-template.md for query and information collection, and @references/writing-guide.md for citation-friendly structure, multi-version writing, and compliance review.
GEO Optimizer Advanced Features
- Multi-engine research across Doubao, Kimi, DeepSeek, Yuanbao, Metaso, Perplexity, and general web search substitutes.
- Citation-source deduplication and ranking based on citation frequency, platform weight, domain relevance, and publishability.
- Creator-center and submission-model analysis that favors outlets the user can actually access.
- One-pass information collection that presents all missing SEO inputs together instead of repeatedly interrupting the user.
- Platform-aware content transformation that adapts one source brief into distinct editorial formats without simple word-for-word duplication.
- GEO-oriented document architecture with TL;DR conclusions, structured headings, explicit facts, source dates, definitions, and FAQ sections designed for extraction.
- Built-in compliance safeguards for advertising restrictions, soft-advertising disclosures, medical and financial red lines, and changing platform rules.
- Browser integration through
web-accessfor rendered pages, direct verification, and authenticated publishing workflows. - WeChat Official Account draft handoff through
wechat-oa-draft-push. - Optional dynamic delegation to
fore-vip-autofor paid topic execution, payment, write-back, refund, and step-return workflows. - Post-publication measurement loop that reruns the original user questions and explains citation gaps caused by content quality, authority accumulation, or insufficient query coverage.
SKILL.md
Loading
Related Openclaw Skills
Product Benchmark · FÓRMULA 100K
Runs deep, multi-tool product and competitor benchmarks in any niche, delivering an HTML report, a 10-dimension scoring matrix, and a 30/60/90 action plan.
Cazador de Lead Magnets F100K
Cazador de Lead Magnets F100K turns real audience pain points into validated, tangible lead magnets with an actionable production brief.
Date Publisher
Date Publisher transforms one approved showme/v1 dating card into controlled, platform-specific public profile drafts and authorized publications.
Douyin Video Copy Extract
A Node.js video-to-text skill that turns Douyin links or local video files into polished copy, summaries, subtitles, scripts, and other prompt-driven content.
Landing de Alta Conversión (Tripwire) — FÓRMULA 100K
Generates a single self-contained, high-conversion tripwire landing page (index.html) with countdown, price anchoring, bonus stacking, guarantee, FAQ and creator bio, ready to paste into GHL or deploy to Vercel/Netlify.
Roadmap Producto F100K
A Backward Design roadmap engine that turns any product promise into a phased customer journey plus a full creator production plan, offer stack, and 90-day financial projection.
Featured*
