Skip to content

AI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web - then synthesizes a grounded summary

refactor(query): extract SOCIAL_NOISE and VIRAL_NOISE shared sets (#437)

Ilia Alshanetsky9fff90b62bce
diff --git a/skills/last30days/scripts/lib/bluesky.py b/skills/last30days/scripts/lib/bluesky.py
index f75d399e..603872f4 100644
--- a/skills/last30days/scripts/lib/bluesky.py
+++ b/skills/last30days/scripts/lib/bluesky.py
@@ -158,14 +158,8 @@ def _reset_session_cache() -> None:
def _extract_core_subject(topic: str) -> str:
"""Extract core subject from verbose query for Bluesky search."""
- from .query import extract_core_subject
- _BSKY_NOISE = frozenset({
- 'best', 'top', 'good', 'great', 'awesome',
- 'latest', 'new', 'news', 'update', 'updates',
- 'trending', 'hottest', 'popular', 'viral',
- 'practices', 'features', 'recommendations', 'advice',
- })
- return extract_core_subject(topic, noise=_BSKY_NOISE)
+ from .query import SOCIAL_NOISE, extract_core_subject
+ return extract_core_subject(topic, noise=SOCIAL_NOISE)
def _parse_date(item: Dict[str, Any]) -> Optional[str]:
diff --git a/skills/last30days/scripts/lib/instagram.py b/skills/last30days/scripts/lib/instagram.py
index b3372446..39eca7df 100644
--- a/skills/last30days/scripts/lib/instagram.py
+++ b/skills/last30days/scripts/lib/instagram.py
@@ -69,17 +69,8 @@ def _resolve_transcript_timeout(
def _extract_core_subject(topic: str) -> str:
"""Extract core subject from verbose query for Instagram search."""
- from .query import extract_core_subject
- _INSTAGRAM_NOISE = frozenset({
- 'best', 'top', 'good', 'great', 'awesome', 'killer',
- 'latest', 'new', 'news', 'update', 'updates',
- 'trending', 'hottest', 'popular', 'viral',
- 'practices', 'features',
- 'recommendations', 'advice',
- 'prompt', 'prompts', 'prompting',
- 'methods', 'strategies', 'approaches',
- })
- return extract_core_subject(topic, noise=_INSTAGRAM_NOISE)
+ from .query import VIRAL_NOISE, extract_core_subject
+ return extract_core_subject(topic, noise=VIRAL_NOISE)
def _to_hashtag_form(query: str) -> str:
diff --git a/skills/last30days/scripts/lib/pinterest.py b/skills/last30days/scripts/lib/pinterest.py
index 1c589832..08002a65 100644
--- a/skills/last30days/scripts/lib/pinterest.py
+++ b/skills/last30days/scripts/lib/pinterest.py
@@ -27,17 +27,8 @@ from .relevance import token_overlap_relevance as _compute_relevance
def _extract_core_subject(topic: str) -> str:
"""Extract core subject from verbose query for Pinterest search."""
- from .query import extract_core_subject
- _PINTEREST_NOISE = frozenset({
- 'best', 'top', 'good', 'great', 'awesome', 'killer',
- 'latest', 'new', 'news', 'update', 'updates',
- 'trending', 'hottest', 'popular', 'viral',
- 'practices', 'features',
- 'recommendations', 'advice',
- 'prompt', 'prompts', 'prompting',
- 'methods', 'strategies', 'approaches',
- })
- return extract_core_subject(topic, noise=_PINTEREST_NOISE)
+ from .query import VIRAL_NOISE, extract_core_subject
+ return extract_core_subject(topic, noise=VIRAL_NOISE)
def _log(msg: str):
diff --git a/skills/last30days/scripts/lib/query.py b/skills/last30days/scripts/lib/query.py
index 38c76a10..e735e623 100644
--- a/skills/last30days/scripts/lib/query.py
+++ b/skills/last30days/scripts/lib/query.py
@@ -54,6 +54,28 @@ NOISE_WORDS = frozenset({
})
+# Shared noise sets for adapter `_extract_core_subject` wrappers.
+#
+# SOCIAL_NOISE: short-form micro-social platforms (Bluesky, Threads, Truth Social)
+# where research/meta words rarely appear in the body of a post.
+SOCIAL_NOISE = frozenset({
+ 'best', 'top', 'good', 'great', 'awesome',
+ 'latest', 'new', 'news', 'update', 'updates',
+ 'trending', 'hottest', 'popular', 'viral',
+ 'practices', 'features', 'recommendations', 'advice',
+ 'or', 'and',
+})
+
+# VIRAL_NOISE: viral / discovery platforms (TikTok, Instagram, Pinterest) and
+# the base for YouTube. Adds 'killer', the prompt-meta cluster, and the
+# methodology cluster on top of SOCIAL_NOISE.
+VIRAL_NOISE = SOCIAL_NOISE | frozenset({
+ 'killer',
+ 'prompt', 'prompts', 'prompting',
+ 'methods', 'strategies', 'approaches',
+})
+
+
def extract_core_subject(
topic: str,
*,
diff --git a/skills/last30days/scripts/lib/threads.py b/skills/last30days/scripts/lib/threads.py
index c7d2a0bc..628b171f 100644
--- a/skills/last30days/scripts/lib/threads.py
+++ b/skills/last30days/scripts/lib/threads.py
@@ -36,18 +36,12 @@ def _extract_core_subject(topic: str) -> str:
emits "A OR B") return zero. Strip boolean operators and cap to the two
most salient words.
"""
- from .query import extract_core_subject
- _THREADS_NOISE = frozenset({
- 'best', 'top', 'good', 'great', 'awesome',
- 'latest', 'new', 'news', 'update', 'updates',
- 'trending', 'hottest', 'popular', 'viral',
- 'practices', 'features', 'recommendations', 'advice',
- 'or', 'and',
- })
- core = extract_core_subject(topic, noise=_THREADS_NOISE, max_words=2)
+ from .query import SOCIAL_NOISE, extract_core_subject
+ core = extract_core_subject(topic, noise=SOCIAL_NOISE, max_words=2)
return " ".join(core.rstrip("?!.").split()[:2])
+
def _parse_date(item: Dict[str, Any]) -> Optional[str]:
"""Parse date from Threads item to YYYY-MM-DD.
diff --git a/skills/last30days/scripts/lib/tiktok.py b/skills/last30days/scripts/lib/tiktok.py
index 81e7a39b..c35771eb 100644
--- a/skills/last30days/scripts/lib/tiktok.py
+++ b/skills/last30days/scripts/lib/tiktok.py
@@ -31,17 +31,8 @@ from .relevance import token_overlap_relevance as _compute_relevance
def _extract_core_subject(topic: str) -> str:
"""Extract core subject from verbose query for TikTok search."""
- from .query import extract_core_subject
- _TIKTOK_NOISE = frozenset({
- 'best', 'top', 'good', 'great', 'awesome', 'killer',
- 'latest', 'new', 'news', 'update', 'updates',
- 'trending', 'hottest', 'popular', 'viral',
- 'practices', 'features',
- 'recommendations', 'advice',
- 'prompt', 'prompts', 'prompting',
- 'methods', 'strategies', 'approaches',
- })
- return extract_core_subject(topic, noise=_TIKTOK_NOISE)
+ from .query import VIRAL_NOISE, extract_core_subject
+ return extract_core_subject(topic, noise=VIRAL_NOISE)
def expand_tiktok_queries(topic: str, depth: str) -> List[str]:
diff --git a/skills/last30days/scripts/lib/truthsocial.py b/skills/last30days/scripts/lib/truthsocial.py
index 83043e65..c0a338dd 100644
--- a/skills/last30days/scripts/lib/truthsocial.py
+++ b/skills/last30days/scripts/lib/truthsocial.py
@@ -33,14 +33,8 @@ def _strip_html(html: str) -> str:
def _extract_core_subject(topic: str) -> str:
"""Extract core subject from verbose query for Truth Social search."""
- from .query import extract_core_subject
- _TS_NOISE = frozenset({
- 'best', 'top', 'good', 'great', 'awesome',
- 'latest', 'new', 'news', 'update', 'updates',
- 'trending', 'hottest', 'popular', 'viral',
- 'practices', 'features', 'recommendations', 'advice',
- })
- return extract_core_subject(topic, noise=_TS_NOISE)
+ from .query import SOCIAL_NOISE, extract_core_subject
+ return extract_core_subject(topic, noise=SOCIAL_NOISE)
def _parse_date(status: Dict[str, Any]) -> Optional[str]:
diff --git a/skills/last30days/scripts/lib/youtube_yt.py b/skills/last30days/scripts/lib/youtube_yt.py
index cc0b358d..33742d6f 100644
--- a/skills/last30days/scripts/lib/youtube_yt.py
+++ b/skills/last30days/scripts/lib/youtube_yt.py
@@ -217,25 +217,17 @@ def _extract_core_subject(topic: str) -> str:
NOTE: 'tips', 'tricks', 'tutorial', 'guide', 'review', 'reviews'
are intentionally KEPT — they're YouTube content types that improve search.
"""
- from .query import extract_core_subject
- # YouTube-specific noise set: smaller than default, keeps content-type words
- _YT_NOISE = frozenset({
- 'best', 'top', 'good', 'great', 'awesome', 'killer',
- 'latest', 'new', 'news', 'update', 'updates',
- 'trending', 'hottest', 'popular', 'viral',
- 'practices', 'features',
- 'recommendations', 'advice',
- 'prompt', 'prompts', 'prompting',
- 'methods', 'strategies', 'approaches',
- # Temporal/meta words — planner generates these but they don't
- # appear in YouTube titles, so strip them for better search.
+ from .query import VIRAL_NOISE, extract_core_subject
+ # YouTube extends VIRAL_NOISE with temporal/meta words the planner emits
+ # that don't appear in YouTube titles (months, recent year tokens, etc.).
+ _YT_EXTRA = frozenset({
'last', 'days', 'recent', 'recently', 'month', 'week',
'january', 'february', 'march', 'april', 'may', 'june',
'july', 'august', 'september', 'october', 'november', 'december',
'2025', '2026', '2027',
'music', 'public', 'appearances', 'developments', 'discussions', 'coverage',
})
- return extract_core_subject(topic, noise=_YT_NOISE)
+ return extract_core_subject(topic, noise=VIRAL_NOISE | _YT_EXTRA)
def expand_youtube_queries(topic: str, depth: str) -> List[str]:
diff --git a/tests/test_query.py b/tests/test_query.py
index a737fcbd..b8048550 100644
--- a/tests/test_query.py
+++ b/tests/test_query.py
@@ -4,6 +4,8 @@ import unittest
from lib.query import (
NOISE_WORDS,
+ SOCIAL_NOISE,
+ VIRAL_NOISE,
extract_compound_terms,
extract_core_subject,
infer_query_intent,
@@ -128,6 +130,59 @@ class TestNoiseWordsCompleteness(unittest.TestCase):
+class TestSharedAdapterNoiseSets(unittest.TestCase):
+ """Pin SOCIAL_NOISE / VIRAL_NOISE so adapters can rely on stable membership.
+
+ Bluesky, Threads, Truth Social use SOCIAL_NOISE.
+ TikTok, Instagram, Pinterest use VIRAL_NOISE.
+ YouTube extends VIRAL_NOISE with temporal/meta tokens (asserted in its
+ own adapter test).
+ """
+
+ def test_social_noise_membership(self):
+ # Words shared with the historical _BSKY_NOISE / _TS_NOISE / _THREADS_NOISE.
+ expected = {
+ 'best', 'top', 'good', 'great', 'awesome',
+ 'latest', 'new', 'news', 'update', 'updates',
+ 'trending', 'hottest', 'popular', 'viral',
+ 'practices', 'features', 'recommendations', 'advice',
+ 'or', 'and',
+ }
+ self.assertEqual(set(SOCIAL_NOISE), expected)
+
+ def test_viral_noise_is_social_superset(self):
+ self.assertTrue(SOCIAL_NOISE.issubset(VIRAL_NOISE))
+ # The extra words VIRAL adds on top of SOCIAL: the historical
+ # tiktok / instagram / pinterest delta.
+ delta = VIRAL_NOISE - SOCIAL_NOISE
+ self.assertEqual(
+ delta,
+ {'killer', 'prompt', 'prompts', 'prompting',
+ 'methods', 'strategies', 'approaches'},
+ )
+
+ def test_extract_core_subject_with_social_noise(self):
+ # Sanity: a Bluesky-style query strips through SOCIAL_NOISE.
+ result = extract_core_subject(
+ "best new Claude Code update",
+ noise=SOCIAL_NOISE,
+ )
+ self.assertEqual(result, "claude code")
+
+ def test_extract_core_subject_with_viral_noise(self):
+ # Viral set strips 'killer' and the prompt cluster.
+ result = extract_core_subject(
+ "killer prompting strategies for React",
+ noise=VIRAL_NOISE,
+ )
+ # 'for' is in the default NOISE_WORDS path but VIRAL_NOISE alone
+ # doesn't include articles/prepositions; extract_core_subject
+ # falls back to original when nothing survives, so allow 'for'.
+ self.assertIn("react", result)
+ self.assertNotIn("killer", result)
+ self.assertNotIn("prompting", result)
+
+
class TestInferQueryIntent(unittest.TestCase):
"""Tests for infer_query_intent() — the shared canonical classifier.