FAQ

FAQ

1. What is the default admin URL and account?

• Admin URL: /geo_admin/
• Default admin username: admin
• Default admin password: password

After first login, change the admin password immediately. For source deployments, make sure Laravel APP_KEY has been generated with php artisan key:generate or by the Docker init service.

2. Is Docker required?

No.

You can:

• use Docker Compose for postgres + redis + init + app + queue + scheduler + reverb
• or run PHP and PostgreSQL locally

If the goal is to get started quickly and predictably, Docker is the safer option.

3. Is PostgreSQL required at runtime?

Yes. The current public runtime database is PostgreSQL. The repository does not include production databases or sample business data.

4. Why are image libraries, knowledge files, and article data not included?

Because those are runtime or business assets, such as image-library contents, raw knowledge-base files, generated articles, logs, and backups. The public repository only includes source code and configuration templates.

5. How do I connect an AI model?

Go to:

AI Configurator -> AI Model Settings

Fill in the API URL, model ID, API key, and model type. GEOFlow supports OpenAI-compatible providers and native Gemini chat / embedding routes.

For knowledge-base RAG, also configure an embedding model and set it as the default embedding model.

6. Why does a knowledge base show chunks but zero vectors?

This usually means no usable embedding model has been configured yet.

GEOFlow can still split the knowledge document into chunks, but vector writing requires:

• an enabled model whose type is embedding
• a valid API key
• a provider endpoint that supports embeddings
• the model set as the default embedding model

After configuring it, re-save, re-upload, or refresh the knowledge base so vectors can be written.

7. Does semantic chunking always require an LLM?

No.

The default structured rule chunking is enough for most documents. LLM semantic planning is useful for long documents, complex structures, or cases where semantic completeness matters. If LLM planning fails, GEOFlow falls back to rule chunking.

8. What is the content-generation pipeline?

The core flow is:

1. Configure models and prompts
2. Prepare knowledge, title, keyword, image, and author libraries
3. Create a task with generation limits, publishing cadence, and publication scope
4. Let the scheduler enqueue it
5. Let the worker execute generation
6. Route content through draft / review / publish
7. Deliver the result to the frontend
8. If channels are selected, enqueue distribution

9. Which target sites does Distribution Management support?

Current primary channel types are:

• GEOFlow Agent: best for fresh target sites. Download a preconfigured PHP package with static pages, sitemap, llms.txt, and TXT maps.
• WordPress REST: best for existing WordPress sites. Use Application Password authentication to publish, update, delete posts, and sync media, categories, and tags.

10. What is the difference between “local only” and “channels only”?

• Local only: publish only on the GEOFlow local frontend. Channel selection is disabled and no distribution job is created.
• Channels only: keep a management copy in GEOFlow while the public content is generated on selected target channels.

11. Why do old article images fail after migration?

Older records may store paths like /uploads/...; the Laravel rewrite serves uploaded public files through /storage/uploads/....

Current code normalizes these paths at render time, but the actual files must still exist under storage/app/public/uploads/... and php artisan storage:link must be valid.

12. Where do Analytics logs come from?

Log analysis reads parseable access logs or imported logs, then extracts visit trends, top content, status codes, source types, and AI crawlers. If no real logs exist yet, sample logs can be used to validate charts and filters.

13. Is there CLI or companion Skill support?

Yes.

• Companion skill repository: yaojingang/yao-geo-skills
• Current key skills:
  • yao-geoflow-template
  • yao-geoflow-design
  • yao-geoflow-cli

14. When is GEOFlow a good fit?

It is a good fit for:

• independent GEO websites
• GEO sub-channels inside official sites
• internal GEO content operations
• knowledge-base-driven content production
• multi-section or multi-site content operations
• WordPress or lightweight target-site distribution

It is not a good fit if the goal is mass-producing low-value pages.

15. Which pages should I read first?

Recommended order:

1. Getting Started
2. What Is GEOFlow
3. Model Setup Guide
4. Knowledge Chunking and RAG
5. Distribution Management and Target Sites
6. Analytics and Logs