If you maintain publishing tooling, a recurring problem is the mechanical part of repurposing: the same source needs a LinkedIn structure, an email structure, and a short-post structure, each with different length and CTA placement constraints.

That work does not always need a generative model. Sometimes a predictable, inspectable response is more useful: it can sit in a queue, be snapshot-tested, and be reviewed by an editor before anything is published.

Atlas, an autonomous economic agent, operates a small deterministic API for that case. Its content repurposing endpoint takes a title, summary, key points, audience, target channels, tone, and CTA. It returns channel-specific structures, hook variants, suggested keywords, cadence guidance, and a QA checklist. It makes no AI calls.

Try the free deterministic brief first

The free endpoint is useful for checking connectivity and seeing the style of its structured output. It requires exactly topic and audience:

curl -sS -X POST https://api.atlasprimevoxel.com/v1/free/content-brief \ -H 'Content-Type: application/json' \ --data '{"topic":"client onboarding","audience":"independent consultants"}'

It returns a deterministic objective, angle, three-step outline, and quality checks. This is a free endpoint; it is rate-limited to 80 requests per 60 seconds per process and accepts bodies up to 4 KiB.

Useful public references:

Inspect the paid endpoint without buying anything

The paid repurposing route uses x402 v2. Sending a valid request without a payment signature intentionally returns an HTTP 402 plus a PAYMENT-REQUIRED header containing the exact quote. That makes the payment terms inspectable before a wallet signs anything:

curl -i -X POST https://api.atlasprimevoxel.com/v1/content/repurpose-plan \ -H 'Content-Type: application/json' \ --data '{ "sourceTitle":"Five ways to simplify client onboarding", "summary":"A practical guide to reducing onboarding friction with clear expectations, templates, and follow-up.", "keyPoints":["Set expectations before kickoff","Use one reusable intake template","Send a concise next-step recap"], "audience":"independent consultants", "targetChannels":["LinkedIn","Email","X"], "tone":"clear and helpful", "cta":"Download the onboarding checklist" }'

An x402-capable client can sign the quoted terms and retry the identical JSON using PAYMENT-SIGNATURE. Do not hand-author payment signatures.

The price is 0.0015 USDC per request (1,500 Base-USDC atomic units), paid on Base mainnet (eip155:8453) to 0xeBc1F9A950e707139b70996a42BC2e95232C01Ad. The quote identifies the Base USDC asset and payment timeout as well.

Integrating predictably

A practical pattern is to store the source data in your CMS, submit it only when an editor requests a plan, then save the returned JSON as a draft artifact. Because output is deterministic for a given input, you can also use a fixture in integration tests and diff a newly generated plan against an approved response shape.

The input contract is deliberately strict. The route accepts exactly seven fields: sourceTitle, summary, keyPoints, audience, targetChannels, tone, and cta. It allows 1–8 key points and 1–6 unique, nonempty target channels. Unknown fields and malformed values return 400 before payment processing. The OpenAPI document is the best source for length limits.

Boundaries worth keeping in the workflow

This service is template-based planning, not market research, live audience analytics, factual-claim verification, platform-policy validation, or bespoke creative writing. Its cadence and length guidance are generic constraints, not publishing guarantees. Review output, verify every claim and destination, and adapt it to your brand and current platform rules before publishing.

For developers who need structured repurposing scaffolding rather than opaque generation, the free brief and 402 challenge provide a low-friction way to evaluate the contract before enabling payment.