 Lukas Goldschmidt
|
5339e4c514
feat: dashboard config page, dedup article identity module
|
hace 6 días |
|
Lukas Goldschmidt
|
b22882c580
feat: article_identity module, site_config DB table, debug_dedup tool
|
hace 6 días |
|
Lukas Goldschmidt
|
28f4322c94
feat: seen_articles table + dual-signal clustering + lower title threshold
|
hace 6 días |
|
Lukas Goldschmidt
|
8fe316db24
debug scripts
|
hace 6 días |
|
Lukas Goldschmidt
|
2c049a1c7e
fix wipe.sh: proper .env sourcing, inline Python, respect NEWS_MCP_DB_PATH
|
hace 1 semana |
|
Lukas Goldschmidt
|
ea3bc2c757
fix wipe.sh: source .env properly, not via xargs export
|
hace 1 semana |
|
Lukas Goldschmidt
|
0d116bc74d
add wipe.sh: source .venv + .env, clear all but feed_state
|
hace 1 semana |
|
Lukas Goldschmidt
|
14dbabaad7
add script: clear all news data but keep feed_state rows
|
hace 1 semana |
|
Lukas Goldschmidt
|
f7f08dc990
seed feed_state from .env at startup — INSERT OR IGNORE, leave existing alone
|
hace 1 semana |
|
Lukas Goldschmidt
|
764aac573e
simpler: don't auto-disable feeds removed from .env — leave DB config alone
|
hace 1 semana |
|
Lukas Goldschmidt
|
7f5176281a
fix: track disabled feeds, reset their item count to 0
|
hace 1 semana |
|
Lukas Goldschmidt
|
2ae7bd92e4
fix: feed dedup hash must not include timestamp — only title+url
|
hace 1 semana |
|
Lukas Goldschmidt
|
b3d915c73d
fix: keyword arg for _cluster_is_within_age_window
|
hace 1 semana |
|
Lukas Goldschmidt
|
2670ed9d44
refactor: rewrite polling loop as ClusterPoller class with per-cycle stats
|
hace 1 semana |
|
Lukas Goldschmidt
|
fab4b5ec31
fix: keep cluster_id stable across cycles — never recompute after creation
|
hace 1 semana |
|
Lukas Goldschmidt
|
c3cc8103fe
fix: skip LLM enrichment for clusters that already have entities and keywords
|
hace 1 semana |
|
Lukas Goldschmidt
|
bb2b345be3
fix: prune on payload_ts (event time) + pre-filter articles older than retention
|
hace 1 semana |
|
Lukas Goldschmidt
|
3b3b8ced25
docs: add NEWS_LLM_CONCURRENCY and NEWS_LLM_RATE_LIMIT env vars to README
|
hace 1 semana |
|
Lukas Goldschmidt
|
2b778f45c7
feat: per-provider LLM rate limiter (token bucket)
|
hace 1 semana |
|
Lukas Goldschmidt
|
065421f71f
fix: preserve enriched_at through sanitize_cluster_payload
|
hace 1 semana |
|
Lukas Goldschmidt
|
2db56b7dc0
poller llm calls reduced
|
hace 1 semana |
|
Lukas Goldschmidt
|
0e2119d549
prompt
|
hace 1 semana |
|
Lukas Goldschmidt
|
8813368e83
clamp future timestamps to now on ingest + remove stale prompt test
|
hace 1 semana |
|
Lukas Goldschmidt
|
a907f7bb81
prompt
|
hace 1 semana |
|
Lukas Goldschmidt
|
45094bbb5b
prompt
|
hace 1 semana |
|
Lukas Goldschmidt
|
535f0bef64
extraction prompt again
|
hace 1 semana |
|
Lukas Goldschmidt
|
b73e04cd73
extraction prompt
|
hace 1 semana |
|
Lukas Goldschmidt
|
e77a2e6e3e
prompt, keywords filter, emerging topics - related entities
|
hace 1 semana |
|
Lukas Goldschmidt
|
7981f483f0
fix: cross-topic dedup pass to merge duplicate clusters from different feeds
|
hace 1 semana |
|
Lukas Goldschmidt
|
4ed086e30c
fix: deduplicate related_keywords against related_entities in emerging topics
|
hace 1 semana |
