mcpField
I
instagram-seo

instagram-seo

github.com/social-media-skillsMarketing & growth210763 installsUpdated 2w ago
Repository

About instagram-seo

An Instagram SEO skill that researches relevant queries and writes captions designed for keyword search. It also advises on profile labels and visual or spoken keyword signals, while distinguishing these recommendations from what WoopSocial publishes.

Key features

  1. 01Researches real search queries
  2. 02Builds a 3–5 keyword cluster
  3. 03Writes search-first captions
  4. 04Advises profile labels and alt text
  5. 05Specifies visual and spoken keywords

Use cases

  1. 01Prepare a keyword-focused Instagram caption
  2. 02Align a Reel's text and opening audio with its caption
  3. 03Advise profile fields for niche search discovery

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/social-media-skills/skills --skill instagram-seo

After installing, type/instagram-seoin Claude Code to confirm it loaded.

Bundled files

References and scripts installed along with SKILL.md.

evals/
evals.json
references/
instagram-seo-2026-reality.md
keyword-research-and-surfaces.md
scope-and-connections.md
the-match-framework.md

Details

From the project README

instagram-seo

The IG keyword-search layer — research the queries and write the keyworded caption (and advise the profile labels + visual signals) that make a post rank in Instagram search. The agent researches + writes the search-first caption + specs the on-screen/spoken keyword + advises the profile/alt text; WoopSocial publishes the caption; the Name field, handle, bio, alt text, category, and auto-captions are native/ advised, and on-screen text/audio come from other tools/the creator.

The POV: Instagram is a keyword search engine — keywords replaced hashtags

Since late 2024 (official in 2026), IG shifted from hashtag-based discovery to keyword-based search — hashtags are deprioritized to 3–5 topic labels, and a large and growing share of discovery/new followers now starts from search (verify-quarterly; never quote a precise share). It's a closed system (no external crawl/backlinks) but Google + Bing index IG captions. Search runs semantic analysis → metadata matching (Name field/bio/caption keyword cluster) → visual recognition (Computer Vision checks the alt text + visual match the caption — a mismatch lowers ranking). WoopSocial publishes the caption; the profile labels, alt text, on-screen text, audio, and settings are native/ advised/other-tools. Pairs with caption-writer (the keyword goes in the searchable caption).

Read these first

  1. brand-profile — niche/voice.
  2. the post topic/destination.

The framework: MATCH

(Depth: references/the-match-framework.md.)

  • M — Mine the real queries: scan top Reels' captions/on-screen text + competitor Name fields + Google "People Also Ask"; build a 3–5 keyword cluster. Method only; never fabricate volume.
  • A — Anchor the caption to search: lead the first sentence / ~125 chars with the primary keyword (natural long-tail); weave the cluster; 3–5 hashtag labels — the publishable surface (content.text).
  • T — Tag the profile labels: Name field ("Name | Niche"), keyworded handle, indexable bio (niche/offer/ location/CTA). Profile-level — native/advised.
  • C — Confirm the visual matches: alt text + on-screen text + first-3-sec spoken keyword reinforce and match the caption (Computer Vision); auto-captions on; remove other-platform watermarks. On-screen → image/video tool; script → reels-script; alt text → native.
  • H — Harvest over time across search + Explore: search = keyword-driven, Explore = behaviour-driven; optimize for saves/shares/watch-time; compounds over 90–120 days; no stuffing / irrelevant tags.

The reality (verify-quarterly)

Keyword-based since late 2024 (official 2026); hashtags 3–5 labels; a growing share of discovery from search; closed system but Google/Bing index captions; 3-step search (semantic → metadata cluster → Computer Vision alt-text/ visual match); surfaces (Name field high weight, handle, indexable bio, search-first captions [primary keyword first sentence/125 chars, 150–300 words], alt text [must match], on-screen text transcribed, audio transcribed [first-3-sec keyword], auto-captions lift watch time + are indexed, category/location, 3–5 cluster); search vs Explore; engagement (saves/shares/watch-time/replays > likes; completion); Reels deprioritized (watermarks/ low-res); the Raw Content Revolution (human-made, retention/sends); research method + slow-burn timeline (2–4 wk / 30–60 day / 90–120 day): references/instagram-seo-2026-reality.md. The research method, the surface map (publishable vs native vs other-tools), the caption template, the must-match checklist, the publish block + worked examples: references/keyword-research-and-surfaces.md.

Honest scope (never violate)

  • The agent researches keywords (advises the method, never fabricates volume), writes the search-first caption (content.text, primary keyword first, no stuffing, 3–5 hashtag labels), specs the on-screen text + first-3-sec spoken keyword, and advises Name field/handle/bio/alt text/category.
  • WoopSocial publishes the IG post's caption (POST/REEL/STORY via content.text).
  • WoopSocial does NOT: set Name field/handle/bio (profile — native), alt text (native/at-post), category/location (native), toggle auto-captions (native), read IG Search/Insights (native), render the on-screen text (image/video tool), or record the spoken audio (creator/reels-script).
  • No keyword stuffing (suppressed); no irrelevant hashtags (hurt credibility); the caption must MATCH the visual (Computer Vision — no misleading); remove other-platform watermarks; never fabricate volume or rankings. (Scope, distinctions + connections: references/scope-and-connections.md.)

Distinct from its siblings (route correctly)

instagram-seo (this) = IG-specific keyword search (Name field, closed-system + Google-indexed, Computer Vision caption-visual match, first-3-sec spoken keyword) · social-seo = the cross-cutting framework · pinterest-seo = Pinterest (boards/Trends/Rich Pins, an open web-indexed system) · hashtag-strategy = deprioritized to 3–5 (they pair: tags label, keywords discover) · caption-writer = the craft (pairs) · reels-script = the Reel script.

Where this connects

Reads first: brand-profile + the post topic. Pairs with: caption-writer (the caption), hashtag-strategy (the 3–5 labels), reels-script (the first-3-sec spoken keyword), ideogram/the video tools (the on-screen text overlay). Routes to/from: instagram-growth (the strategy), social-seo/ ai-search-optimization (the cross-platform + AI search picture), content-recycling (re-cut a winning Reel into a keyworded carousel). Publishes via: scheduling-and-queue → WoopSocial (the caption; POST/REEL/ STORY) + platform-specs-and-validation. Measures with: native IG Insights + analytics-and-reporting (profile visits from search, saves, non-follower reach). Profile, alt text, settings + IG Search stay native.

Definition of done

Real queries mined via top-Reel scans + competitor Name fields + Google "People Also Ask" into a 3–5 keyword cluster (method advised, no fabricated volume); a search-first caption with the primary keyword in the first sentence (natural long-tail, no stuffing, 3–5 hashtag labels) written as the publishable surface; the Name field ("Name | Niche"), handle, bio, and alt text advised; the on-screen text + first-3-sec spoken keyword specced (to reels-script + an image/video tool) and confirmed to MATCH the caption with auto-captions on and no watermark; the caption published via WoopSocial while profile/alt-text/category/auto-captions/IG-Search stay native; no keyword stuffing, no irrelevant tags, no caption-visual mismatch, nothing fabricated; correctly distinguished from social-seo and pinterest-seo and paired with caption-writer.

Comments

Markdown supported