Analytics Skills

Conversation and usage analytics, ad-account intelligence, SEO and answer-engine visibility, security posture and product signals

24 skills

• Ad Account Auditor

  Structured full-account audit in strict order (tracking first, then structure, bidding fit, computed search-term waste, assets, settings, landing pages) — weighted score, prioritized fix list, estimated monthly waste; unverifiable areas stated as unverified, never passed

  View skill →

• Agent Interaction Analytics

  Query aggregate, organization-scoped interaction statistics across every chatbot in this organization: per-day usage report (conversations, messages, sessions, tokens), daily tool activity, tool usage, and widget-context breakdown. Aggregates only, never raw conversation content.

  View skill →

• Answer-Engine Optimization (GEO)

  Visibility inside AI assistants, where a buyer sees one answer naming two or three vendors instead of a page of results: tracks buyer questions, asks them of each engine, and records whether the brand was named, at what position, which competitors were named instead and what was cited (share of answer). Also audits answer-readiness — AI crawler access, llms.txt, entity schema, question-shaped structure, extractable answers, freshness, concrete claims — and turns gaps into dated evidence for the product signal ledger

  View skill →

• Budget Pacing Engine

  Cross-client morning sweep: spend vs each client's budget of record, month-end projection (Google 30.4× + 2× overdelivery, Meta weekly flex), conversion-lag-adjusted anomaly scoring with flatline detection, correction drafts — observes, never operates

  View skill →

• Client Report Composer

  Self-assembling client report packs: pacing vs budget first, per-platform + blended KPIs with prior-period deltas, tracking-status caveats, recent-activity context and a narrative skeleton — delivery only through the proposal queue

  View skill →

• Competitor Ad Intel

  Standing watches on public ad libraries per client watchlist: active-ad counts, format mixes, recency and dated example snapshots with mandatory provenance (snapshot URL + date) — never copy, always counter

  View skill →

• Competitor Price Watch

  Monitor competitor catalogs/prices from PUBLIC pages (schema.org), match to the shop's products, and report price position + undercut alerts. Public sources only; findings for your own team — never signals or coordinates with a competitor.

  View skill →

• Compliance Report Generator

  Generate comprehensive compliance reports from learner enrollment, certification, and audit data. Uses AI to structure data into executive summaries, section breakdowns, and custom fields.

  View skill →

• Conversation Analytics

  Analyze conversation data: sentiment trends, topic extraction, response quality metrics, and session summaries.

  View skill →

• Creative Fatigue Monitor

  Per-creative baselines and lifecycle staging Healthy → Early Warning → Fatigued → Dead: hook rate vs own baseline for video, CTR decay + rising frequency for statics; the replacement queue drives the creative brief loop with ~48h lead time

  View skill →

• Demographic Estimation

  Privacy-first demographic estimation from camera with explicit consent. Client-side face-api.js and server-side vision LLM modes.

  View skill →

• Extension Proposal Engine

  The product evolution backlog and its human gate: evidence-bound extension proposals (new vertical OS, new skill, integration, improvement, pricing, go-to-market) ranked on an open formula, red-team critiques recorded with the proposal, and a decision queue only a named human can clear — an agent calling the decide tool is refused by design

  View skill →

• Launch QA Gate

  Pre-launch checklist executor with binary verdicts recorded per campaign (live URL+UTM validation, tracking/budget/settings evidence, client sign-off) — the campaign launch refuses to execute without a recorded PASS

  View skill →

• Market Gap Miner

  Find assortment gaps vs competitor coverage, score "stock next" recommendations by demand evidence (unserved requests, trend, competitor coverage), and plan seasonal preparation. Evidence-graded; analyses only.

  View skill →

• Media Plan Forecaster

  Versioned media plan of record per client-month: channel splits that must add up, forecast ranges from the client's own trailing data with confidence stated, scenario comparison with the diminishing-returns caveat, mandatory change note per revision

  View skill →

• Pentest Orchestrator

  Authorized penetration-testing lifecycle: validate the signed scope, authorize/refuse/human-gate a proposed action, build a remediation report. Out-of-scope, DoS/destructive and third-party actions refused by construction; active exploitation and intrusive production tests are non-waivable human gates.

  View skill →

• Product Signal Ledger

  The evidence base behind every product decision: dated, sourced signals from usage telemetry, support threads, interviews, win/loss reviews, churn post-mortems, market scans and competitive intelligence, clustered into named patterns with a strength, a breadth and an evidence score. Tenant-walled — a signal records that a pattern exists, never the customer content behind it

  View skill →

• Search Term Miner

  Search-term hygiene: n-gram waste analysis, converting-query promotion drafts, negative-list drafts classified against client brand terms, PMax brand-leak audit — outputs are drafts for the proposal queue, never direct mutations

  View skill →

• Security Posture Scan

  Non-intrusive security assessment of an AUTHORIZED target: security headers (CSP strength, HSTS, clickjacking protection, referrer/permissions policy), cookie flags on session-shaped cookies, HTTPS enforcement, permissive CORS, information disclosure (version banners, stack traces) and a fixed four-entry check for well-known exposed files. Refuses anything outside an active scope; no fuzzing, brute force, payloads or exploitation

  View skill →

• Security Scanner

  Authorized, non-intrusive vulnerability scanning of the shop's OWN assets: security headers, TLS, dependency/CVE audit, posture report. Every scan is bounded by the owner-signed in-scope register — out-of-scope refused. Observes and reports; never exploits.

  View skill →

• SEO Audit

  Crawl/rank readiness for a URL: indexability first (a noindex page fails before anything else matters), title and meta description, heading hierarchy, canonical, language and viewport, image alt coverage, thin content, internal linking, structured data and sharing tags — every finding with a severity and the concrete fix. SSRF-safe fetching; measures, never edits

  View skill →

• Site Audit Engine

  Audit the storefront's technical and on-page SEO health (title/meta/H1/canonical/hreflang, images missing alt, mobile viewport, mixed content, broken pages) into prioritized findings. Findings only; never modifies the site.

  View skill →

• Tracking Sentinel

  Measurement-integrity watch: onboarding access/tracking checklist (pixel↔CAPI dedup, Enhanced Conversions for Leads, Consent Mode v2, offline imports), conversion-flatline detection, page tag probes, and the per-client green/amber/red scorecard that gates optimization — red never auto-clears

  View skill →

• Visual Charts Generator

  Generate visual charts and graphs from data. Supports bar, line, pie, scatter, area, and combo charts as PNG, SVG, or interactive HTML.

  View skill →