What is trend_pulse?
trend_pulse is a Hooklayer MCP tool that returns what is peaking in short-form video right now for a given niche. Three rising opportunities (format / hook / style / topic) with growth percentages, plus two saturated patterns to avoid. Sources: Google Trends RSS, YouTube Data API velocity, Reddit hot, and an internal 100K-video corpus. 12-hour cache per niche.
Format trends, not news topics. trend_pulse surfaces what video FORMAT is rising ("Sports Finance Breakdown +178.4%", "Teaser Trailer Format +241.6%") rather than "trending news topics" — because content velocity in short-form is structural, not topical. A format works for months; a news topic dies in 48 hours.
Decimal-precision growth. Percentages are emitted with one decimal place and explicitly forbidden from rounding to multiples of 5 or 10. "+47.2%" reads as an estimate from noisy data; "+50%" reads as fabricated. The forced precision is the credibility signal.
Per-entry provenance. Each rising/saturated entry carries sources[] (which data feeds surfaced it), signal_window (last 7 days for rising, last 30 for saturated), and signal_strength 0-1. Top-level provenance carries cache_age_hours and cache_status: fresh | stale so agents can decide whether to refetch.
Inputs & outputs
Endpoint: POST /api/v1/trends
Inputs
nichestringNiche filter (e.g. "Finance & Business", "Beauty & Skincare"). Omit for broadly applicable trends.
Output fields
rising3 rising trends with title, type, growth (decimal %), competition, suggestion, sources[], signal_window
saturated2 saturated patterns with status (cooling | saturated | overused)
generated_atWhen the underlying digest was last refreshed (12-hour TTL per niche)
provenancedata_sources[], cache_age_hours, cache_status: fresh | stale, signal_window
qualitylevel (full | partial | degraded if cache > 24h)
cURL
curl -X GET "https://hooklayer.dev/api/v1/trends?niche=Finance+%26+Business" \
-H "Authorization: Bearer hl_live_..."Example prompts
Paste any of these into Claude Desktop (with Hooklayer connected) to see the live response.
Niche trend pulse with provenance
Use Hooklayer trend_pulse for the "Finance & Business" niche. Show me the 3 rising trends with their growth percentages (decimal, not rounded), sources[], and signal_window for each. Also cite the top-level provenance.cache_status so I know if I'm looking at fresh data.Expected output: Returns 3 rising trends like Sports Finance Breakdown +178.4% / Teaser Trailer +241.6% / Behind-Scenes +118.7% with sources[] listing Google Trends + YouTube velocity + Reddit hot. cache_status: fresh or stale.
Cross-niche trend comparison
Call Hooklayer trend_pulse three times — once for "Beauty & Skincare", once for "Fitness & Health", once for "Tech & Coding." Compare the rising trends across the three niches. Which format is rising in ALL three? Which is niche-specific?Expected output: Demonstrates format-level patterns that cross niches (e.g. Teaser Trailer rising everywhere) vs niche-specific ones (e.g. "GRWM" beauty-only).
Verify before content planning
I'm planning 12 TikTok videos for my finance niche over the next month. First, call Hooklayer trend_pulse for "Finance & Business" and flag any trend with cache_status: stale — I don't want to plan around 24-hour-old data. Then suggest which 3 rising formats I should rotate through.Expected output: Surfaces cache freshness as a content-planning signal. Recommends 3 formats with competition levels for cadence planning.
Frequently asked
How is this different from Google Trends?
Google Trends shows search-query velocity. trend_pulse shows short-form VIDEO FORMAT velocity — synthesized from Google Trends + YouTube view rates + Reddit thread frequency + an internal 100K-video corpus. The result is "what format is winning right now" rather than "what topic is people searching for." Different layer of the same intent stack.
Why decimal percentages instead of round numbers?
Round numbers (+180%) read as fabricated. Real estimates from noisy data look like +178.4% or +241.6%. The prompt explicitly forbids rounding to multiples of 5 or 10. The decimal is the credibility signal — a way to make the estimate inspectable rather than authoritative-sounding-but-unverifiable.
What does signal_window mean?
The time horizon over which the trend was measured. Rising trends use the last 7 days (fast-moving). Saturated patterns use the last 30 days (slower decay). Each entry includes its own signal_window so agents can weight recent signals more heavily than long-decay ones.
Can I get trends for a specific country or region?
Not in v1. trend_pulse aggregates global signals weighted toward English-language US/UK content. Per-region trend filtering ships in v2 alongside region-specific MCP routing. Workaround: pass a niche string that implies region (e.g. "UK Finance" vs "Finance & Business") — the model adjusts but doesn't guarantee region-exclusive data.
What's cache_status fresh vs stale?
Cache TTL is 12 hours per niche. fresh = response was generated within the last 12 hours. stale = the cached response is between 12 and 24 hours old (you're seeing it because no one has triggered a refresh yet). >24 hours triggers a refetch automatically. Agents should weight stale data slightly lower or call again to force a refresh.
Why only 3 rising and 2 saturated?
Decision-tool focus. A list of 15 trends is harder to act on than a curated 3+2. The constraint forces the upstream synthesis to pick the highest-signal items and discard noise. If you need broader breadth, call trend_pulse on multiple niches and combine — that's what most agents do.
Related tools
Try trend_pulse in 30 seconds.
100 free credits at signup. No card. Works in Claude Desktop, Cursor, n8n, or any MCP client.