sinapse-ai 9.3.0 → 9.5.0

package/squads/squad-content/knowledge-base/signal-intelligence-v2.md

@@ -0,0 +1,234 @@
+# Signal Intelligence v2 — 2026 Sources + Classification Canon
+
+> **Status:** Canon (squad-content v2.0) — ADDITIVE companion to `signal-detection-methods.md` v1
+> **Owner:** signal-intelligence (Radar) — primary; content-orqx (Nexus) — curator
+> **Created:** 2026-04-12 for Story fw-v2.4
+> **Explicit relationship:** Does NOT replace `signal-detection-methods.md`. v1 remains canonical for foundational definitions. v2 adds the 2026 source landscape, the signal classification taxonomy, routing rules and a quality-scoring model that did not exist in v1.
+
+---
+
+## 1. Why v2 Exists
+
+The v1 KB (`signal-detection-methods.md`) was written assuming a pre-Grok / pre-Apify-ubiquity landscape. Since then:
+
+- **X (Twitter) API v2** has become the dominant surface for public discourse, with tier changes that reshape what's scrapable.
+- **TikTok Research API** became available for commercial signal detection and is now the #1 cultural-signal source for sub-25 audiences.
+- **Apify actors** became the pragmatic alternative when APIs are closed — a rented actor marketplace for scraping any social surface.
+- **Grok on X** became a first-order summarizer of live discourse, shifting the signal-detection workflow.
+- **Substack** became a newsletter-first signal source for professional niches.
+- **Reddit API** re-opened for commercial use under new pricing.
+- **Google Trends + YouTube Data API v3** remain stable but are now cross-referenced against the above rather than primary.
+
+v1 still defines what a signal IS and the 7 signal types. v2 defines where signals LIVE in 2026 and how to score them.
+
+---
+
+## 2. Current-State Signal Source Catalog
+
+### 2.1 Tier A — First-party research APIs (highest trust)
+
+| Source | Access | Rate | Best for | Cost |
+|--------|--------|------|---------|------|
+| X API v2 (Basic/Pro/Enterprise) | OAuth 2.0, Basic tier ~$100/mo | ~10k posts/mo Basic | Real-time discourse, influencer signals, thread analysis, Grok summaries | Paid |
+| TikTok Research API | Academic/commercial app approval | Batch hashtag queries | Cultural signals, sub-25 behavior, meme velocity, trend lifecycle | Free with approval |
+| YouTube Data API v3 | Google Cloud key | 10k units/day free | Long-form trend signals, creator signals, search intent behind video queries | Free tier + paid |
+| Reddit API (official) | OAuth 2.0, commercial tier | Paid tiers for commercial | Niche community signals, question mining, "authentic voice" samples | Paid for commercial |
+| Google Trends API (pytrends + official) | Public, rate-limited | Rolling | Search intent velocity, regional variation, seasonality | Free |
+
+### 2.2 Tier B — Scraping / Actor-based sources
+
+| Source | Access | Best for | Legal/ToS note |
+|--------|--------|---------|----------------|
+| Apify actors (Instagram, TikTok, LinkedIn, X, etc.) | Apify platform pay-per-run | Whatever is not open via official API | Respect robots.txt + ToS, document legal basis per actor |
+| Substack RSS feeds | Public RSS | Professional niche signals, newsletter thesis drift | Public RSS is safe |
+| Newsletter aggregators (Feedly, Inoreader) | API keys | Curated professional signals by editor | OK |
+| Podcast RSS + transcription (Whisper) | Public RSS + local Whisper | Long-form thesis signals before they hit text | Respect IP/quotation rules |
+
+### 2.3 Tier C — Curated human-intelligence
+
+| Source | Why | Cadence |
+|--------|-----|---------|
+| Exploding Topics | Trend pre-emergence detection | Weekly digest |
+| SparkToro | Audience intelligence + zero-click research | On demand |
+| BuzzSumo | Evergreen scoring + viral performance | On demand |
+| Semrush Trends | SEO-backed topic signals | Rolling |
+| Parse.ly Currents | Publisher-side trend intelligence | Daily |
+| Stacked Marketer / Morning Brew / The Hustle / Superpath | Human-curated daily scan | Daily |
+
+### 2.4 Tier D — Proprietary client signals (highest leverage)
+
+| Source | Why it matters |
+|--------|---------------|
+| Client DMs, comments, reviews, support tickets | Authentic audience voice — highest-leverage signal, hardest to find |
+| Client analytics (Parse.ly, GA4, native platform analytics) | Performance signals closing the loop back to retrofeed |
+| Sales team transcripts (Gong, Fireflies) | What the market is saying with buying intent |
+| Customer interview transcripts (Whisper) | Pain language, vocabulary, objections |
+
+**Rule of thumb:** Tier D signals > Tier A > Tier B > Tier C for relevance. Tier C and A are broader-coverage; Tier D is narrower-but-hotter.
+
+---
+
+## 3. Signal Classification Taxonomy
+
+Every detected signal MUST be classified into exactly one of these 5 categories. Classification determines routing to Phase 2 (AI Draft) and the editorial pillar mapping.
+
+### 3.1 Opportunity Window
+
+**Definition:** A short-lived condition where content on a specific topic will over-perform because attention is concentrated.
+**Velocity:** Hours to 7 days.
+**Example:** Apple releases a new dev tool at WWDC — window to publish "what this means for SaaS founders" is 48-72h before commodity hot-takes flood the channel.
+**Routing:** Phase 1 → Phase 2 immediately, high urgency, human-edit fast-track.
+**Failure mode:** Late-to-the-window = commodity content.
+
+### 3.2 Trend Seed
+
+**Definition:** Early sign of a topic gaining velocity before it becomes mainstream. Detected pre-inflection.
+**Velocity:** Weeks to months.
+**Example:** A term starts appearing in 3+ unrelated niche newsletters over 2 weeks. Not yet on Google Trends.
+**Routing:** Phase 1 → editorial calendar insertion, build a pillar-aligned series.
+**Failure mode:** Calling a trend too early or forcing a seed that won't grow.
+
+### 3.3 Crisis Alert
+
+**Definition:** A brand-relevant event that forces a response — competitor misstep, industry controversy, client-adjacent scandal, regulatory change.
+**Velocity:** Immediate (hours).
+**Example:** A major competitor gets caught in a data breach — opportunity for a thoughtful "here is how we think about data stewardship" piece.
+**Routing:** Phase 1 → escalate to client immediately, Phase 2 requires human-heavy review (high reputational stakes).
+**Failure mode:** Opportunistic copy that reads as vulture-like; miss the window of relevance.
+
+### 3.4 Category Shift
+
+**Definition:** A slow but durable change in how the category thinks, talks or operates. Signals arrive from multiple unrelated sources over time.
+**Velocity:** Months to quarters.
+**Example:** The shift from "content marketing" to "audience-first content" in 2018-2021. You can see it in retrospect; the leverage is seeing it 6 months before the category does.
+**Routing:** Phase 1 → manifesto piece + editorial pillar reshape, high authority play.
+**Failure mode:** Chasing every language shift as if it were a category shift.
+
+### 3.5 Competitor Move
+
+**Definition:** A direct, attributable action by a known competitor — launch, campaign, messaging shift, hiring pattern, product change.
+**Velocity:** Days to weeks.
+**Example:** Competitor launches a new tier; opportunity to reframe your own positioning.
+**Routing:** Phase 1 → analyst pass (`benchmark-against-competitors`) → Phase 2 strategic response content.
+**Failure mode:** Obsessively responding to competitors instead of leading the category.
+
+---
+
+## 4. Signal-to-Pillar Routing Rules
+
+Every signal must be routed to an editorial pillar before it enters Phase 2. Unrouted signals are archived.
+
+### 4.1 Routing Algorithm
+
+```
+1. Classify signal into one of 5 categories (§3)
+2. Match signal to existing editorial pillars via semantic overlap
+3. If match → route to pillar, schedule in Phase 2 queue
+4. If no match → check if the signal is a Category Shift (rare but possible)
+   - YES → escalate to editorial-strategist for pillar review
+   - NO → archive with "no pillar fit" tag
+5. If match spans multiple pillars → assign primary pillar + cross-reference secondary
+```
+
+### 4.2 Routing Priority Matrix
+
+| Signal Category | Priority | Route Directly To |
+|----------------|----------|-------------------|
+| Crisis Alert | P0 | content-orqx → editorial-strategist (escalation) |
+| Opportunity Window | P1 | Phase 2 fast-track, content-engineer |
+| Competitor Move | P2 | content-analyst benchmark, then content-engineer |
+| Trend Seed | P3 | editorial-strategist for calendar insertion |
+| Category Shift | P3 | editorial-strategist for pillar reshape consideration |
+
+---
+
+## 5. Signal Quality Scoring
+
+Every signal, before entering Phase 2, is scored 1-5 on 5 dimensions. A signal must score ≥ 3.0 average to enter Phase 2.
+
+| Dimension | 1 | 5 |
+|-----------|---|---|
+| **Relevance** to pillar/client | Unrelated | Perfect pillar fit |
+| **Novelty** vs prior content | Covered recently | Fresh angle |
+| **Evidence** backing the signal | 1 weak source | 3+ independent sources |
+| **Actionability** for target audience | No clear reader action | Clear reader takeaway |
+| **Window** still open | Closed/saturated | Open and ahead of curve |
+
+Final score = average of 5 dimensions. Below 3.0 → archive. 3.0-4.0 → Phase 2 standard. 4.0-5.0 → Phase 2 priority + optional cross-platform batch.
+
+---
+
+## 6. Source Monitoring Cadence
+
+| Tier | Source | Cadence |
+|------|--------|---------|
+| A | X API v2 | Continuous (webhooks/streams where possible) |
+| A | TikTok Research API | Daily batch query per pillar |
+| A | YouTube Data API | Daily batch |
+| A | Reddit API | Twice daily |
+| A | Google Trends | Twice daily |
+| B | Apify actors | Daily (rotating by cost) |
+| B | Substack RSS | Continuous |
+| C | Exploding Topics / SparkToro / BuzzSumo | Weekly |
+| C | Newsletters | Daily at fixed time |
+| D | Client DMs / analytics | Continuous |
+| D | Sales transcripts | Weekly batch |
+
+---
+
+## 7. Anti-Patterns (v2 additions over v1)
+
+- **API-only monitoring** — missing Tier D (client-proprietary) signals is the most expensive blind spot.
+- **Grok as sole summarizer** — powerful but opinionated; cross-reference with raw X posts.
+- **Scraping without legal basis** — Tier B Apify usage must document basis per actor.
+- **Treating Crisis Alerts as Opportunity Windows** — speed without care damages brand.
+- **Category Shift false positives** — most language shifts are not category shifts; require multi-source, multi-month evidence.
+- **Signals without routing** — unrouted signals clutter the pipeline and never become content.
+- **Ignoring Tier D in favor of Tier A** — public-web signals are the same ones everyone else sees.
+
+---
+
+## 8. Handoff to Phase 2 (AI Draft)
+
+When a signal passes quality scoring and routing, the hand-off package to Phase 2 contains:
+
+```
+signal:
+  id: sig-{yyyymmdd}-{n}
+  category: opportunity_window | trend_seed | crisis_alert | category_shift | competitor_move
+  source_tier: A | B | C | D
+  source_name: e.g., "X API v2"
+  evidence:
+    - url: ...
+      excerpt: ...
+    - url: ...
+      excerpt: ...
+  pillar_primary: ...
+  pillar_secondary: [...]
+  quality_score: 4.2
+  window_open_until: 2026-04-15T18:00Z
+  suggested_thesis: "..."
+  suggested_format: ["carousel", "thread"]
+```
+
+This is the structured input that `content-engineer` (Arc) consumes at Phase 2 of `ai-native-content-loop.md`.
+
+---
+
+## 9. Relationship to v1 KB
+
+| Topic | v1 (signal-detection-methods.md) | v2 (this KB) |
+|-------|----------------------------------|--------------|
+| What is a signal | DEFINITIVE | Referenced |
+| 7 signal types | DEFINITIVE | Referenced |
+| Source catalog | Stale (2026 update) | **DEFINITIVE** |
+| Classification taxonomy | Not covered | **DEFINITIVE** |
+| Routing rules | Not covered | **DEFINITIVE** |
+| Quality scoring | Not covered | **DEFINITIVE** |
+| Handoff format | Not covered | **DEFINITIVE** |
+
+---
+
+## Change Log
+
+- 2026-04-12 — v1.0 authored as part of Story fw-v2.4 (squad-content v2.0 upgrade). Additive to v1 signal-detection-methods.md — v1 is NOT modified.