A daily digest with a memory for quality
4 ranked items across 4 topics from 5 sources. You see public source URLs, headlines, coarse relevance, and source health, never reviewer notes or the scoring prompt.
Per-topic coverage
Strong, limited, or absent coverage for each configured topic in the batch.
Ranked items
Items in rank order with their topic, coarse relevance, and public source, links open the already-public source.
- 1Anthropic ships agent skills framework updateAgentsRelevance 94%anthropic.com
- 2EU AI Act governance guidance for high-risk systemsGovernanceRelevance 88%digital-strategy.ec.europa.eu
- 3New open-source developer tool for tracing LLM callsDeveloper toolsRelevance 81%github.com
- 4Survey on AI in K-12 education toolingEducationRelevance 73%arxiv.org
Source-health scorecard
Useful signal, noisy or duplicate sources, failed fetches, and empty sources, the inputs the digest is judged on.
3
Useful signal
1
Noisy / duplicate
1
Failing fetch
0
Empty
- Healthy —anthropic.com6 items0 failsUseful signal
- Healthy —digital-strategy.ec.europa.eu3 items0 failsUseful signal
- Degraded —github.com2 items1 failsNoisy
- Healthy —arxiv.org4 items0 failsUseful signal
- Failing —src_blog_x (unreachable)0 items3 failsFailing
Judge and reader feedback
Coarse feedback counts only, free-text feedback is never shown. Feedback creates candidates, it does not silently mutate the template.
Learning timeline
Recent judge reviews, reader feedback, candidate promotions, and source quarantines.
Judge review×12
4d ago
Reader feedback×5
4d ago
Candidate promoted×1
5d ago
Source quarantined×1
6d ago
Why this is valuable to us
The digest is not just a feed, it is a live signal on source health, topic demand, a governance watchlist, knowledge-base freshness, and what the template should learn next.
- Source health
- 3 of 5 sources healthy
- Topic demand
- 2 topics with strong coverage
- Governance watchlist
- 2 sources on the watchlist
- Template-learning signal
- 17 feedback signals folded in