Hashtag Strategist

Intelligent social-media hashtag strategy microservice: suggest a platform-tuned mix of reach/niche/branded hashtags for a topic (Instagram, TikTok, X/Twitter, LinkedIn, YouTube, Pinterest, Threads...

hashtag-strategist

Intelligent social-media hashtag strategy microservice. Turns a topic into a platform-tuned mix of reach / niche / branded hashtags, scores individual tags for competitiveness vs discoverability, screens for banned/shadowban-risk tags, and enforces per-platform count limits.

Every tool has a deterministic heuristic core that runs fully offline (word-shape scoring, reach tiers, curated banned list, platform tables). suggest additionally has an optional LLM path that enriches topic→tag ideation and silently falls back to the heuristic when no model is reachable — results are tagged { mode: 'heuristic' | 'llm' }.

Tools

Tool	Args	Returns
`suggest`	`{ topic, platform?, count?, options?:{ seed, brand, mix, allowLLM } }`	hashtag set split into reach/niche/branded tiers
`score`	`{ hashtag }`	`{ competitiveness, reach, discoverability, tier, sizeBand, banned, advice }`
`bannedCheck`	`{ hashtag }` or `{ hashtags:[...] }`	per-tag safe/banned/broken/caution flags
`mixStrategy`	`{ topic? , hashtags?, platform?, count?, options?:{ mix, brand } }`	balanced reach/mid/niche plan + breakdown
`byPlatform`	`{ platform? }`	count limits, sweet-spot, default mix, formatting notes (one or all)

Platforms: instagram, tiktok, twitter/x, linkedin, youtube, pinterest, threads, facebook.

Example

import pack from './index.js';
await pack.adapters.suggest({ topic: 'vegan meal prep', platform: 'instagram', count: 11 });
// → { mode, platform:'instagram', hashtags:[...], tiers:{ reach, niche, branded }, detail:[...] }

pack.adapters.score({ hashtag: '#love' });
// → { competitiveness: 96, tier:'reach', sizeBand:'mega (>5M posts…)', advice:'…' }

pack.adapters.bannedCheck({ hashtags: ['#follow4follow', '#veganmealprep', '#123'] });
// → flags follow4follow (banned) and 123 (all-numeric/broken)

Heuristic model

Offline we cannot query a live post count, so competitiveness/reach is estimated from tag shape — the same signals a growth marketer eyeballs: short generic single words ⇒ mega/saturated (reach tier), longer multi-word/specific tags ⇒ small/niche tier with high discoverability. The banned list is a curated snapshot of community-flagged / permanently-broken tags plus structural rules (all-numeric, over-long, double underscores).

DRY boundaries

Not seo (keyword density, meta tags, SERP snippets) — this is social hashtag strategy.
Not nlp (tokenization/sentiment) — reused conceptually only.
Self-contained: the only cross-file import is ../_shared/llm.js for the optional AI path.