AI agent skill that researches any topic across Reddit, X, YouTube, HN, Polymarket, and the web - then synthesizes a grounded summary
Python ai-promptsai-skillblueskyclaudeclaude-codeclawhubdeep-researchhackernewsinstagramopenclawpolymarketrecencyredditresearchsocial-mediatiktoktrendstwitterweb-searchyoutube
"""U1 - nominate stage: river/listing candidate discovery.
Covers the two behaviors the nominate stage adds over the old inline sweep:
a ``keyword_gate`` toggle (domain scoping vs global trending) and
fault-tolerant per-source failure recording that never raises.
"""
from unittest import mock
from lib import pipeline, reddit_listing, schema
def _plan(domain: str, sources: list[str], subreddits: list[str] | None = None) -> schema.DiscoveryPlan:
return schema.DiscoveryPlan(
domain=domain,
category=None,
subreddits=subreddits or ["all"],
sources=sources,
)
def test_fetch_discovery_source_reddit_gate_filters_off_domain():
"""With the keyword gate on, off-domain listing items are dropped; with it
off (global trending), the feed's own hot ranking is kept verbatim."""
plan = _plan("AI agents", ["reddit"])
payload = {
"items": [
{"title": "New AI agents framework launched", "selftext": ""},
{"title": "Gardening tips for spring", "selftext": ""},
],
"errors": [],
}
with mock.patch.object(reddit_listing, "fetch_discovery_listings", return_value=payload):
gated, _ = pipeline._fetch_discovery_source(
"reddit", plan,
from_date="2026-06-10", to_date="2026-07-10",
depth="default", mock=False, config={}, keyword_gate=True,
)
ungated, _ = pipeline._fetch_discovery_source(
"reddit", plan,
from_date="2026-06-10", to_date="2026-07-10",
depth="default", mock=False, config={}, keyword_gate=False,
)
assert [item["title"] for item in gated] == ["New AI agents framework launched"]
assert len(ungated) == 2
def test_nominate_candidates_threads_keyword_gate():
"""nominate_candidates forwards its keyword_gate to each source fetch, so a
global (no-domain) run really does disable the gate."""
plan = _plan("", ["reddit"])
seen: dict[str, bool] = {}
def fake_fetch(source, plan, *, from_date, to_date, depth, mock, config, keyword_gate=True, warnings=None):
seen["keyword_gate"] = keyword_gate
return [], None
with mock.patch.object(pipeline, "_fetch_discovery_source", side_effect=fake_fetch):
pipeline.nominate_candidates(
plan,
from_date="2026-06-10", to_date="2026-07-10",
depth="default", mock=False, config={}, lookback_days=30,
keyword_gate=False,
)
assert seen["keyword_gate"] is False
def test_nominate_candidates_records_source_failure_without_raising():
"""One dead feed is recorded on the bundle as a failure; the surviving feed
still yields candidates, and the call never raises."""
plan = _plan("AI agents", ["reddit", "hackernews"])
def fake_fetch(source, plan, *, from_date, to_date, depth, mock, config, keyword_gate=True, warnings=None):
if source == "hackernews":
raise TimeoutError("hn listing timed out")
return pipeline._mock_discovery_items(source, plan.domain, to_date), None
with mock.patch.object(pipeline, "_fetch_discovery_source", side_effect=fake_fetch):
bundle = pipeline.nominate_candidates(
plan,
from_date="2026-06-10", to_date="2026-07-10",
depth="default", mock=False, config={}, lookback_days=30,
)
# Surviving source produced candidates.
assert bundle.items_by_source.get("reddit")
# Dead source recorded as a failure, not silently dropped or raised.
hackernews = bundle.source_status["hackernews"]
assert hackernews.state not in (schema.NO_RESULTS,)
assert "ok" != hackernews.state