
by Orthogonal
Find, score, and enrich Twitter/X influencers for any brand in minutes
Suits developers who need an agent to assemble a creator outreach shortlist, then pass it to other Orthogonal skills such as Google Sheets. It covers only Twitter/X, not Instagram, TikTok or YouTube, and its SKILL.md has been unchanged since 27 February 2026.
Running the full discovery-to-contact pipeline (find, score, enrich email and LinkedIn) in a single agent turn without managing separate API keys for each data source.
Orchestrating 5 sub-APIs in a single SKILL.md with a scoring model is a meaningful workflow advantage; replicating it manually takes hours. But dedicated platforms like SparkToro have better data quality.
Main trade-off
Exa cannot search Twitter profiles directly, so discovery depends entirely on third-party curated list pages. Influencers not covered in listicles will be systematically missed.
Billing is per endpoint call on your Orthogonal balance, with no monthly platform fee. Orthogonal says each endpoint has a catalog price, fixed or request-dependent. The SKILL.md workflow is free and open source, so the spend is the Exa, Scrape Creators, Fiber, Hunter and Tomba calls it triggers. The vendor published no per-run total when checked on 26 October 2026.
npm install -g @orth/cli && orth skills add orthogonal/find-twitter-influencers
Agent follows SKILL.md instructions, calling orth run exa /search, orth run scrapecreators /v1/twitter/profile, and enrichment endpoints in parallel via the orth CLI
The permissions this connector asks for, as its own documentation states them.
6 actions an agent can call. Open one for its parameters and an example.
Example call
orth run brand-dev /v1/brand/retrieve-by-name --query 'name=Acme Corp'
Parameters
| Parameter | Type | Needed | Description |
|---|---|---|---|
| name | string | Optional | Company name to resolve (use this or domain). |
| domain | string | Optional | Company domain to resolve directly. |
Returns
| Field | Type | Description |
|---|---|---|
| domain | string | Resolved company domain. |
| industry | string | Detected industry / category. |
Example call
orth run exa /search --body '{"query": "best fintech Twitter accounts to follow", "numResults": 10, "contents": {"text": {"maxCharacters": 5000}}}'Parameters
| Parameter | Type | Needed | Description |
|---|---|---|---|
| query | string | Required | Search query, e.g. 'best {niche} Twitter accounts to follow'. |
| numResults | integerdefault 10 | Optional | Results per query. |
Returns
| Field | Type | Description |
|---|---|---|
| results | array | Matched pages with title, url and extracted text content. |
Example call
orth run exa /findSimilar --body '{"url": "https://example.com/top-fintech-twitter-influencers", "numResults": 5, "contents": {"text": true}}'Parameters
| Parameter | Type | Needed | Description |
|---|---|---|---|
| url | string | Required | URL of a known-good curated list page. |
| numResults | integerdefault 5 | Optional | Number of similar pages to return. |
Returns
| Field | Type | Description |
|---|---|---|
| results | array | Similar pages with title, url and text content. |
Example call
orth run scrapecreators /v1/twitter/profile --query 'handle=examplehandle' orth run scrapecreators /v1/twitter/user-tweets --query 'handle=examplehandle'
Parameters
| Parameter | Type | Needed | Description |
|---|---|---|---|
| handle | string | Required | Twitter/X handle without the @ symbol. |
Returns
| Field | Type | Description |
|---|---|---|
| followers_count | integer | Current follower count. |
| legacy.favorite_count | integer | Per-tweet like count used in the engagement-rate calculation. |
Example call
orth run fiber /v1/kitchen-sink/person --body '{"profileIdentifier": "https://linkedin.com/in/janesmith"}'
orth run hunter /v2/email-finder --query 'domain=janesmithcreative.com' --query 'first_name=Jane' --query 'last_name=Smith'Parameters
| Parameter | Type | Needed | Description |
|---|---|---|---|
| profileIdentifier | string | Optional | LinkedIn URL for Fiber kitchen-sink lookup (preferred, best match rate). |
| domain | string | Optional | Company domain for Hunter's name+domain fallback. |
Returns
| Field | Type | Description |
|---|---|---|
| string | Best-match email address, when found. | |
| linkedin_url | string | LinkedIn profile URL, when found. |
Example call
orth run -X POST nyne /person/newsfeed -d '{"social_media_url": "https://x.com/TARGET"}'
orth run sixtyfour /enrich-lead --body '{"lead_info": {"first_name": "Jane", "linkedin_url": "https://linkedin.com/in/janesmith"}}'Parameters
| Parameter | Type | Needed | Description |
|---|---|---|---|
| social_media_url | string | Optional | X/Twitter profile URL for Nyne's async newsfeed fetch. |
| lead_info | object | Optional | Known identifiers (name, LinkedIn URL) passed to Sixtyfour for deep enrichment. |
Returns
| Field | Type | Description |
|---|---|---|
| request_id | string | Nyne async job id, polled via a follow-up GET. |
| work_email | string | Sixtyfour's best-match work email, when found. |
Find Twitter Influencers is an Orthogonal SKILL.md workflow that runs 3 parallel discovery strategies (Exa search, similar-page expansion, Fiber natural-language search) to surface Twitter/X creators for a brand, then ranks them with a 5-factor composite scoring model before returning a contact-enriched shortlist.
3-Strategy Discovery Pipeline
Runs Exa curated-list search across 5+ query variations spanning core and adjacent niches, widens coverage by pulling pages similar to the strongest listicle finds, and layers in Fiber natural-language profile search, all in parallel before filtering.
Scrape Creators Profile and Tweet Enrichment
Fetches structured Twitter profiles for all candidates (followers, bio, verification) and then tweets for the top 40 candidates, calculating average likes, retweets, replies, and engagement rate from the nested legacy.favorite_count fields.
5-Factor Composite Scoring
Ranks candidates using a weighted model: relevance to the brand (40%), engagement rate (25%), follower count log-scaled (15%), original content quality (10%), and estimated audience alignment (10%).
Contact Enrichment Waterfall
Finds LinkedIn URLs via Exa site:linkedin.com searches, then enriches contacts via Fiber kitchen-sink (LinkedIn URL as profileIdentifier), Hunter email-finder (domain + name), and Tomba LinkedIn-to-email in that priority order.
Hard Filter Application
Removes candidates with fewer than 10,000 followers, no tweets in the last 30 days, empty or off-niche bios, suspended accounts, and accounts that have migrated to Mastodon, Bluesky, or Threads.
Ranked Markdown Table Output
Delivers a ranked table of 20 influencers by default (configurable) with full plain-text URLs for Twitter profile, top tweet, email, and LinkedIn, plus a summary of candidate pool size, filter drop-off, and niche coverage.
Optional Deep-Dive Enrichment
For any influencer needing more detail, an optional step layers in Nyne for an async full tweet-history newsfeed and Sixtyfour for AI-powered lead research (work email, audience size, collab history) beyond the default ranked table.
Building an influencer outreach list for a product launch
An agent takes a brand name and niche, runs the full discovery-to-contact pipeline, and returns 20 ranked influencers with email and LinkedIn ready for an outreach draft.
Chaining influencer research into a full outreach pipeline
Because the skill runs inside Claude Code or Cursor, its output can feed directly into other Orthogonal skills like Gmail or Google Sheets to draft and log outreach without leaving the agent.
Repeatable research across multiple client niches
Agencies run the same documented 9-step pipeline against different client niches, getting a consistent ranked table instead of manual Twitter browsing each time.
The nearest AI agent skills on Hokai for the same job.
| Skill | Works with | Connects by | Checked |
|---|---|---|---|
| Claude · Cursor | MCP | 26 Oct | |
| -- | MCP | 17 Nov | |
| Claude · Cursor | REST API | 11 Oct | |
| Claude · ChatGPT · Gemini | REST API | 11 Oct |
The skill file is free and open source; what you pay for is the API calls it makes on Orthogonal, where each endpoint (Exa search, Scrape Creators profile and tweet fetches, Fiber, Hunter, Tomba) carries its own catalog price. Orthogonal lists no monthly platform subscription and takes dollar credits by card. The vendor does not publish a per-run total, so the cost of one run depends on how many candidates and adjacent niches you search. Checked against Orthogonal's pricing page on 26 October 2026.
Reading and copying the SKILL.md from the orthogonal-sh/skills repository costs nothing and needs no account. Running it does need an Orthogonal key with a funded balance, because every Exa, Scrape Creators and enrichment call is billed per endpoint. A free starting credit was listed on an earlier version of this page, but it could not be confirmed on Orthogonal's current pages on 26 October 2026, so check the dashboard after signing up.
SparkToro is a dedicated audience-research platform that finds accounts by what a brand's customers already follow, a different signal than keyword-and-listicle discovery. Modash is an influencer database covering Twitter, Instagram, TikTok and YouTube with a searchable filter interface built for non-technical marketers. HypeAuditor adds bot-detection and audience-quality scoring this skill's pipeline does not include. Pick this skill when the research needs to run inside an agent alongside other automated steps.
SparkToro is a standalone subscription product that ranks accounts by shared audience overlap, a different signal than this skill's keyword-and-listicle discovery. Find Twitter Influencers has no subscription of its own; cost follows usage through Orthogonal's per-endpoint pricing. SparkToro ships its own dashboard and saved lists, while this skill returns a table to whichever agent called it, ready to feed outreach drafting or a spreadsheet.
Install the CLI with npm install -g @orth/cli, set your Orthogonal API key, then add the skill with orth skills add orthogonal/find-twitter-influencers so the SKILL.md lands in the agent's skills directory. Open Claude Code or Cursor and ask the agent to find Twitter influencers for a brand, giving it a company name and optionally a niche and result count (the default is 20). The agent runs the pipeline itself and returns a ranked table.
The Find Twitter Influencers skill from Orthogonal is a multi-step workflow packaged as a SKILL.md file that teaches AI agents how to discover, score, and enrich Twitter/X influencers for a given company or product. Rather than a single API call, it runs 3 research strategies in parallel: Exa web search for curated influencer list pages (Exa does not index x.com profiles, so the skill reads listicles and blog posts instead), a similar-page expansion off the strongest list finds, and Fiber natural-language profile search to surface LinkedIn-heavy professionals. From those sources it extracts handles, pulls profiles and engagement metrics through Scrape Creators, scores each candidate, and enriches contact details with Hunter, Tomba and Fiber lookups. The default output is a ranked table with followers, engagement rate, top tweet URL, email and LinkedIn.
The skill lives in the open-source orthogonal-sh/skills GitHub repository, in its own orthogonal-find-twitter-influencers folder. Its SKILL.md was last committed on 27 February 2026, while the repository as a whole last merged a change on 17 June 2026. It is built by Orthogonal, the Y Combinator-backed gateway that announced a $4.3M seed round led by Pantera Capital. Agents invoke the skill by reading the instructions and then calling orth CLI commands such as orth run exa /search and orth run scrapecreators /v1/twitter/profile, all routed through one Orthogonal key, so the agent never manages separate keys for Exa, Scrape Creators, Fiber, Hunter or Tomba. The same pattern powers siblings like Orthogonal Find Leads and Orthogonal Find Skill, and the table output chains naturally into Google Sheets or Google Calendar for follow-up.
The workflow is built for parallelism: profile fetches, tweet fetches, LinkedIn lookups and email enrichment calls are launched in batches. Scoring is a 5-factor composite that weighs relevance, engagement rate, follower count, content quality and audience alignment, and it applies hard filters of at least 10,000 followers (adjustable), a tweet in the last 30 days, and a bio relevant to the niche or an adjacent one. Exa does most of the discovery work, so the quality of the shortlist is capped by what public listicles already cover.
Cost is usage-based. Orthogonal says each endpoint has a price in its API catalog, some fixed and some dependent on the request, that credits are added in dollars by card (agents can also pay through x402, MPP or Nanopayments), and that the platform has no monthly subscription. The SKILL.md itself is free to read and copy. A full run touches many endpoints, so the total depends on candidate volume; check each endpoint's catalog price before a large run.
Who should skip it: marketers who want a point-and-click database with saved lists, audience-overlap data or bot detection will be better served by a dedicated influencer platform, because the skill returns its results to the calling agent and has no dashboard of its own. It also covers Twitter/X only, and it will under-find creators who never appear in a public list. Browse more agent capabilities in the skills directory, or let Smart Match shortlist tools for a specific outreach job.
One change to know about: the hosted skill page the listing used to link to now redirects to the Orthogonal homepage, and the repository README catalog does not list this skill, although its SKILL.md remains in the repo.
Protocol, authentication, permissions and the call contracts on this page are read from Orthogonal's own documentation. The verdict is Hokai's reading of them, not the maintainer's claim.