aggregators/kagi-news/src · coves · Tangled

bretton.dev / coves

A community based topic aggregation platform built on atproto

coves aggregators kagi-news src

12 files

Bretton · fix(kagi): stop trusting the model to echo story IDs in semantic dedup · 2w ago

Merge branch 'feat/kagi-news-aggregator-phase-1'

refactor(kagi-news): switch from HTML scraping to Kagi JSON feed with inline citations

feat(kagi-news): add semantic deduplication via Claude Haiku

coves_client.py

fix(kagi): trim megathread sources to the lexicon's 50-entry cap

fix(kagi): skip null perspective entries instead of losing the story

fix(kagi): stop trusting the model to echo story IDs in semantic dedup

refactor(kagi-news): switch from HTML scraping to Kagi JSON feed with inline citations

richtext_formatter.py

fix(uri): normalize URIs to the atproto uri format on write

refactor(kagi-news): switch from HTML scraping to Kagi JSON feed with inline citations

semantic_dedup.py

fix(kagi): stop trusting the model to echo story IDs in semantic dedup

state_manager.py

fix(kagi): stop trusting the model to echo story IDs in semantic dedup

uri_sanitizer.py

fix(validation): allowlist http/https with a required host for link URIs