Docs

Every Orange Slice capability and provider operation: exact inputs, output schema, price and failure codes. Generated from the API itself, so it is never out of date.

Building an agent? Every page has a markdown twin, and the full index is at /llms.txt.

Capabilities

One outcome, one price, whichever providers it takes.

• ai_extract$0.002, charged per call

  Extracts structured fields from text against a JSON Schema you supply. One credit per call. This is the one to use in a pipe — it returns an object the next command can read, where `ai text` returns prose it cannot.

• ai_text$0.002, charged per call

  Runs a prompt through a fast model and returns text. One credit per call, billed whether or not you like the answer — the model did the work either way. Use `ai text_deep` when the fast model is not good enough; it is the same call on a stronger model at ten credits.

• ai_text_deep$0.02, charged per call

  The same call as `ai text` on a stronger model, at ten credits instead of one. Worth it when the fast model has already produced a wrong answer; not worth it as a default.

• apify_run$0.01 per result, minimum $0.10, up to $10.00 per call

  Runs an Apify actor and returns up to 1,000 dataset rows. Reserves five credits a row from the limit you ask for. For a pay-per-event actor read that as a floor rather than a quote — the real cost is not knowable until the actor has run — and note the hard cap of 20,000 credits, enforced on the run itself, so a runaway is stopped rather than billed.

• browser_run$0.002, charged per call

  Runs JavaScript in a hosted browser and returns whatever it returns. One credit per run. Reach for it when a page needs interaction that `scrape website` cannot do — a login, a click, an infinite scroll.

• company_find_careers_pagefree

  Resolves a company website to its careers page, preferring the underlying ATS board (Greenhouse, Lever, Ashby and friends) over a human-facing page. Free — it fetches pages directly. Pipe it into `search company_scrape_careers_page` to get the jobs themselves.

• company_linkedin$0.004, charged per hit

  Finds a company's LinkedIn page from its name or website, and returns the slug that `search employees` takes. Two credits, billed only when a page is found.

• company_linkedin_enrich$0.002, charged per hit

  Reads a company's record from the B2B database — size, industry, description, location — in one indexed lookup. One credit, billed only when the company is found.

• company_scrape_careers_pagefree

  Reads the open roles off an ATS board (Greenhouse, Lever, Ashby, Workday and eight more) and emits one job per line. Free, and no browser — it uses each provider's own API where there is one. A board outside the supported set returns no jobs with a `status` explaining why.

• company_search$0.002 per result, minimum $0.002, up to $4.00 per call

  Finds companies by firmographics — industry, country, region, city, headcount, size band, founding year, ticker, or domain. Billed one credit per row returned, so `--limit` is also your budget. Emits `companySlug`, which `search employees` reads directly.

• crunchbase_search$0.002 per result, minimum $0.002, up to $0.20 per call

  Find companies on Crunchbase by funding stage, amount, recency, category and location.

• employees$0.002 per result, minimum $0.002, up to $0.20 per call

  Finds people at a company by job title. Billed one credit per row returned, so `--limit` is also your budget. Output field names match what `enrich` reads, so the two pipe together directly.

• ocean_companies$0.03 per result, up to $3.00 per call

  Find companies resembling domains you already win, or filter by country, industry, headcount and ecommerce. Billed 15 credits per company returned.

• ocean_people$0.03 per result, up to $3.00 per call

  Find people by seniority, department and job title at companies matched by lookalike domains or firmographics. Billed 15 credits per person returned.

• parsed_address$0.002, charged per hit

  Splits a free-text address into street, city, state, postal code and country, plus latitude and longitude. Costs 1 credit per address. An address the geocoder cannot match is returned as an error rather than an empty row, and is not billed.

• person_linkedin$0.004, charged per hit

  Finds a person's LinkedIn profile URL from their name plus a company or title. Searches, then has a model check the result against what you supplied, so a returned URL has been sanity-checked. Two credits, billed only when a profile is found.

• person_linkedin_enrich$0.002, charged per hit

  Reads a person's LinkedIn profile from the B2B database — title, company, location, skills — in one indexed lookup rather than a scrape. One credit, billed only when the profile is found.

• person_linkedin_from_email$0.10, charged per hit

  Finds a person's LinkedIn profile URL from an email address, via FullEnrich's reverse lookup. Fifty credits, billed only when a profile is found — twenty-five times the price of the name-based `person_linkedin`, because it is a different and much more expensive lookup. Reach for it when a name search has already failed.

• phone_number$0.50, charged per hit

  Finds a phone number for a person. Costs ten times a work email, so reach for it on high-value prospects rather than a whole list. Billed only when a number is found.

• places$0.01 per result, up to $1.00 per call

  Finds local businesses on Google Maps and returns one row per place, with address, phone, website and rating. Costs 5 credits per place returned — a search that returns 12 places costs 60. Up to 500 credits are held while it runs (100 places, the pinned maximum) and the unused remainder is released. A search that matches nothing is free.

• scrape_website$0.002, charged per hit

  Fetches a page and returns it as markdown, with its links and metadata. One credit, billed only when content actually comes back — an empty scrape is free. Single page: this does not crawl.

• web_batch_search$0.002 per result, up to $0.50 per call

  Run up to 250 web searches in one throttled call. Billed one credit per search that returns results, so empty searches are free.

• web_search$0.002, charged per hit

  Searches the web and emits one result per line. One credit, billed only when the search returns something. This is the discovery primitive — reach for it when you have no company or identifier to start from.

• work_email$0.05, charged per hit

  Finds a work email for a person. Runs a provider waterfall and validates the result, so a returned address has already passed verification. Billed only when an email is found.

Providers

A specific vendor's endpoints, exposed directly.