AI RSS FILTER & COMBINER
A few things worth knowing.
Does this replace my RSS reader?
No. It sits between your sources and the reader you already use.
Can I filter without AI?
Yes. Choose merge-only and optionally add inclusion or exclusion keywords. Keywords can also run before AI filtering. They match words in article titles only, without case sensitivity; summaries and body text are not matched; any exclusion takes priority. With inclusion words, at least one must match. Leave both lists empty to disable this layer. The editor displays the platform's current combined keyword limit. Keyword rejections use no AI checks. Passing articles still require an AI judgment in AI mode, while merge-only uses no AI checks. Both modes remain subject to normal source, new-article, processing-interval and output limits. The system never switches to merge-only when AI fails.
What is one AI check?
One valid KEEP, DROP or UNCERTAIN judgment for an article in one output at a fixed content/rule/process version. Failed technical attempts, reading RSS and manual corrections do not consume another check.
What happens when checks run out?
New AI judgments pause. Existing output remains readable. Filtering never automatically turns off.
Does it fetch full articles?
No. Only titles, summaries and text supplied by RSS or Atom are used. Missing material can lead to uncertainty.
Are feeds private?
Input URLs are treated as private by default. Output URLs are bearer credentials: anyone holding the link can read that output.
Does feedback train a model?
No. Manual corrections are scoped to one output and do not rewrite your rules or train preferences.
Is it a permanent archive?
No. Each output retains a bounded recent window. Raw text, review history and lightweight deduplication records have separate retention periods.