Infinitum

Integrated lazy cat oidc login, or default password 123456 ##Who is suitable for use - People who want to gather multiple RSS sources on one page to read. - People who want to reduce duplication of news, similar events, and interference from low-value content. - People who want to use AI to automatically generate summaries, event groupings, and daily briefings. - I want to retain the ability to manually review rather than hand it over to someone who automatically publishes the algorithm. ##Recommended usage process ### 1. Add RSS sources Enter: ```text /admin ``` Add an RSS address in the source configuration area in the background. It is recommended to start with a small number of high-quality sources, such as technical blogs, project releases, news sites, forum highlights, or security bulletin sources. Focus on checking when adding sources: - Whether the RSS URL can be parsed normally. - Whether the source name is clear. - Whether they are placed in the appropriate grouping. - Whether to enable AI parsing. - Whether to participate in aggregation. If some sources have a lot of content but are noisy, you can keep grabbing, but turn off "Participate in Aggregation" and only view the original entry when needed. ### 2. Configure source grouping Split sources into different groups by topic, for example: - technology news - open source projects - AI - security - product dynamic - personal blog These groups are used for both front-end information flow and back-end filtering. The clearer the grouping, the less effort will be needed for subsequent reading and review. ### 3. Configure AI models Add OpenAI-compatible interfaces to the backend model configuration. The core information that needs to be filled in usually includes: - API Base URL - API Key - model name - whether to enable - concurrency limits If the model is not configured for the time being, Infinitum can still complete basic RSS crawling and display, but AI title translation, summary, event analysis, grouping quality and daily newspaper generation will be limited. ### 4. Check Prompt configuration Prompt is used to control how AI processes content. Infinitum supports Prompt for different purposes, such as: - entry summary - item analysis - Aggregate summary - aggregate match - Daily newspaper generation You can keep the default configuration for the first time. After grabbing a batch of content, adjust Prompt based on the output effect. ### 5. Manually triggered grab After configuring the source and model, manually trigger a grab task in the background. The grab task enters the background queue and is executed by the Worker. Web pages are only responsible for creating tasks, and the real RSS pulling, body catching, AI analysis and event grouping are completed by the background Worker. After triggering, it is recommended to turn on task monitoring and observe: - Whether the task starts. - What stage is the current processing stage. - Whether there are source grabbing failures. - Whether the AI call was successful. - Whether a timeout, network error, or model return exception occurred. ### 6. View the information flow at the front desk After grabbing, return to the home page: The front page is used to read the public information stream. You can filter content by time, source, grouping, and keywords, or view aggregated event cards. Infinitum will mix two types of content: - Aggregate events: When multiple articles describe the same event, they will be merged into a cluster. - Single content: Content that cannot be grouped or is not suitable for grouping will be displayed separately. ### 7. Review and amendment content After entering the background, you can manually maintain the content: - Restores content filtered by rules. - Manually filter low-value content. - Regenerate title translations or summaries. - Reanalyze the entry. - Adjust the relationship between items and event aggregation. - Merge or resume aggregation events. It is recommended to do more reviews at the beginning, as this can help you determine whether source quality, blacklist rules and Prompt need to be adjusted. ### 8. Using AI Daily AI Daily will generate a structured summary from the candidate content of the day, which is suitable for daily centralized review.