Instagram Niche & Performance Intelligence · Open Skill 公開技能
Instagram Niche & Performance Intelligence
Discover Instagram niches and micro-niches, identify what is outperforming, and uncover the content patterns associated with that performance. Discover what works on Instagram — and investigate why. Instagram Niche & Performance Intelligence researches categories, niches, micro-niches, competitors, profiles, and Reels using public evidence. It maps broad markets down to the deepest useful micro-niches, identifies high-performing and outlier Reels, and compares stronger content with reasonable baselines when the available data allows. Instead of simply showing what is popular, it investigates the patterns associated with stronger performance — including hooks, topics, story structures, recurring characters, emotions, formats, duration, visual approaches, Reel covers, and series strategies. Explore Instagram from Category → Niche → Micro-Niche → Profile → Reel → Content DNA, uncover content gaps and emerging opportunities, and continue deeper through evidence-based research paths. What would you like to investigate on Instagram? I can discover niches and micro-niches, research competitors, analyze profiles and Reels, identify outliers, compare stronger-performing content with normal content, and uncover recurring content patterns. Start broad or bring me a specific niche, profile, or Reel.
VTuber
8 1 4
https://d7qlwi7tl2u27.cloudfront.net/skill-covers/prod/223892/2026-09-29/2nnvbhst.png
# Instagram Niche & Performance Intelligence
## When to use
- User wants to discover promising Instagram niches / sub-niches / micro-niches.
- User wants to analyze or validate a specific niche ("analyze senior fitness on Instagram").
- User wants competitor / profile intelligence on an Instagram profile or its competitors.
- User wants Reel intelligence on a specific Reel and why it may be outperforming.
- User wants content performance intelligence on high-performing Reels in a niche and recurring patterns.
- User wants cross-language / cross-market analysis.
- User wants to advance one step of an ongoing Instagram research project.
## Capability model (read first — honest limits, REEL-CENTRIC)
This skill runs on real, currently-available tools. It does NOT have an Instagram Graph API, Instagram Insights API, OAuth access, native hashtag analytics, bulk Instagram historical data, or any private-metric access. Therefore:
- PRIMARY RESEARCH ENGINE = `search_web` (Tavily web search). Use it to discover profiles, Reels, niche discussions, rankings, articles, and community signals in the target language/market.
- REEL METADATA & METRICS = `fetch_video_info` (metadata-only: title, duration, views, likes, comments, channelName, publishedAt, thumbnailUrl, platform) and `download_video` (Instagram, returns title/duration/file size + a temporary download link).
- REEL CONTENT = `download_video` → `recognize_video` (visuals/plot/style/format) and `transcribe_audio` (speech/dialogue). Pass the returned `downloadUrl` verbatim (it may be a `shorturl://…` reference).
- REEL COVER / FIRST FRAME = `recognize_image` on a public thumbnail URL (e.g. `thumbnailUrl` from fetch_video_info).
- REPORTS = `create_document` / `update_document` / `get_document`.
This skill is REEL-CENTRIC. Public Reel metrics (views, likes, comments, date, duration, cover) are the strongest quantitative signal. Follower/following/post counts, bio, full profile history, posting frequency, captions/hashtags, and carousel/image-post metrics are NOT reliably retrievable. Private Instagram Insights (reach, impressions, saves, shares, retention, watch time, profile visits, follows from content, traffic/source, audience demographics) are NEVER available unless the user supplies them.
NEVER fabricate: follower counts, reach, impressions, saves, shares, retention, watch time, audience demographics, traffic/source data, historical metric time-series, native hashtag analytics, or any private metric. If a metric isn't actually retrievable, mark it INSUFFICIENT DATA or ANALYTICAL INFERENCE — never present it as measured fact. Never imply access to unavailable data.
## Reference files (read on demand, don't read all at once)
- `references/research-method.md` — niche trees, query expansion, research depth, time windows, Evidence Ledger, fact-vs-inference, confidence methodology. Read for any research task.
- `references/competition-intelligence.md` — Reel intelligence, competitor discovery, direct vs adjacent, outlier methodology, demand vs supply vs competitive strength, content-gap detection, winning patterns, opportunity mapping, Reel opportunity generator. Read for competitor/profile/opportunity work.
- `references/language-market.md` — language vs market, matrix, scorecards, cross-language gaps, transferability, localization, evidence requirements. Read when language/market comparison is relevant.
- `references/research-transparency.md` — mandatory research transparency & evidence disclosure (RESEARCH & EVIDENCE NOTE, EVIDENCE BASIS, source attribution, claim classification, sample-aware language, cross-market caution, report ending). Read before writing any substantive report.
- `references/external-data.md` — optional External Data / Metric Input Layer (supported formats, activation, provenance schema, conflict handling, derived metrics, transparency integration). Read when the user supplies structured data or asks how to provide it.
- `references/content-performance-int