agentsclimarketplace

Building with influship

Skill Influship/influship-skills/skills/building-with-influship

Official Influship agent skills (build with the API/SDK)

Install
npx -y skills add Influship/influship-skills --skill building-with-influship

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Build, integrate, and debug applications on the Influship creator-discovery API and its official TypeScript SDK (npm `influship`). Use when installing or setting up the Influship SDK, authenticating with an Influship API key, or writing code that calls Influship — semantic creator search, match/fit scoring, lookalike discovery, autocomplete, profile or post lookups, and raw Instagram/YouTube data — and when handling Influship pagination, errors, retries, or credit costs. Also use when calling the Influship REST API directly from a non-TypeScript language (curl, fetch, Python). Not for operating the Influship MCP server (separate skill), and not for general influencer-marketing strategy with no API work.

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

6.3 KB, ~1.4k tokens by cl100k_base, as published. Nobody here has run it

Building with Influship

Influship is a creator-discovery API: semantic search, match scoring, and lookalike discovery across Instagram and YouTube creators, billed by credits. This skill helps you write correct integration code against the TypeScript SDK (npm influship) — and the raw REST API when you are not on TypeScript.

Core principles (read first)

  1. Don't trust training data — verify against the installed package and live docs. The API and SDK evolve. For the version a developer actually has installed, the source of truth is node_modules/influship/api.md and the bundled .d.ts type definitions — these are never stale relative to their code. Never guess SDK bindings (method names, params, response fields); if unsure, read the installed types or fetch a live source. See references/live-sources.md.
  2. Default to the latest SDK version. Don't pin an old version in examples unless the user asks.
  3. Server-side only. The SDK runs on Node 20+, Deno, Bun, Cloudflare Workers, and Vercel Edge — not browsers or React Native (your API key must never ship to a client).
  4. Be credit-aware. Calls cost credits. Bound limit, cache profile lookups, and don't over-paginate. See references/credits-and-cost.md.

Get set up

  1. Get an API key from the developer dashboard: https://developers.influship.com
  2. Install the SDK:
    npm install influship
    
  3. Set INFLUSHIP_API_KEY in your environment (e.g. .env, never committed).
  4. Instantiate the client (server-side):
    import Influship from 'influship';
    
    const client = new Influship({
      apiKey: process.env['INFLUSHIP_API_KEY'], // default; can be omitted
    });
    
    const health = await client.health.check(); // { ok: true, timestamp: ... }
    

Useful env vars: INFLUSHIP_API_KEY, INFLUSHIP_BASE_URL (override base URL), INFLUSHIP_LOG (debug|info|warn|error|off).

Operations cheat-sheet (cached: 2026-05-29 — verify params against installed types)

CallUse it when
client.search.create({ query, limit?, platforms?, creator_kinds?, filters? })Discovery by intent — natural-language creator search. Returns search_id + first page.
client.search.retrieve(id, { cursor?, limit? })Page through the results of a prior search (paginated).
client.creators.match({ creators, intent })Score creators you already have against an intent → good/neutral/avoid + reasons.
client.creators.lookalike({ seeds, filters?, limit? })Expand from seed creators to similar ones (paginated, async-iterable).
client.creators.autocomplete({ q, limit?, platform?, scope? })Resolve a partial name/handle to creator(s).
client.creators.retrieve(id, { include? })Full creator profile (ai_summary, themes, brand_alignment, key_facts; include: ['profiles']).
client.profiles.get(username, { platform })Cached per-platform metrics for one handle.
client.profiles.lookup({ ... })Batch profile lookup (see installed types for exact params).
client.posts.list({ ... })List posts (paginated; see installed types for exact params).
client.raw.instagram.getProfile(username, {...})Live (uncached) Instagram profile scrape.
client.raw.youtube.getChannel / getChannelTranscripts / getTranscript / searchLive YouTube channel, transcripts, and search.
client.health.check()Liveness check (no credits).

Which endpoint when (the #1 confusion):

  • Have an intent but no creators yet → search.create.
  • Have a list of creators and want to know if they fit → creators.match.
  • Have a few great creators and want more like themcreators.lookalike.
  • Have a handle/name and need the canonical creator/profile → autocomplete then creators.retrieve / profiles.get.

Exact, current params and response shapes live in references/operations.md (and ultimately in the installed types / OpenAPI — see references/live-sources.md).

Best-practice pitfalls

  • Pagination: list methods are auto-paginating with for await … of. Don't hand-roll cursors. See references/pagination.md.
  • Errors: catch Influship.APIError subclasses; handle RateLimitError (429) and AuthenticationError (401) explicitly. Retries (2×, backoff) and timeouts (60s) are built in and configurable. See references/errors-and-reliability.md.
  • Credits: see references/credits-and-cost.md before writing loops.
  • Non-TypeScript: call the REST API directly — see references/rest-api.md.

Reading guide

  • Building a real flow end-to-end → references/recipes.md
  • Exact method params / response fields → references/operations.md
  • Paginating large result sets → references/pagination.md
  • Error handling / retries / rate limits → references/errors-and-reliability.md
  • Cost control → references/credits-and-cost.md
  • curl / fetch / Python → references/rest-api.md
  • Anything that might be out of date → references/live-sources.md

Boundary

This skill is for writing integration code. To let an agent call the Influship API live during a session, use the Influship MCP server (a separate skill / setup); this skill does not depend on the MCP being connected.

Verify your work

Before declaring an integration done, run a real call (start with client.health.check(), then the actual endpoint with a tiny limit) and confirm the response shape matches what your code expects.

What ships with it: 7 files

16.3 KB alongside SKILL.md

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.