# Lucidry, Content Refinery for Translation and Naturalization

> Lucidry turns rough drafts, documents and transcripts into clean, natural, publish-ready content, and translates them across languages without losing facts, voice or structure. Use it from the dashboard, the REST API, an MCP client or an SDK.

## Product
- Translate: 10 languages, glossary enforcement, certainty-marker retention, 5-dimension drift tracking
- Naturalize: removes AI language patterns under a bounded change budget
- Clean: encoding, typography and filler cleanup with protected spans locked
- Extract and OCR: PDF, DOCX, HTML and scans to structured text
- Validate: semantic fidelity plus the five-dimension LUCID quality score
- Assemble: DOCX, PDF, HTML and Markdown delivery with checksums

## Developer Integration
- REST API: POST /v1/translate and the v1 job endpoints. https://lucidry.io/docs/api
- MCP server: Content and translation tools over SSE for Claude, Cursor and other MCP clients. https://lucidry.io/docs/mcp
- SDKs: TypeScript and Python clients over the same v1 API. https://lucidry.io/docs/sdk
- MCP connection URL: https://lucidry.base44.app/functions/mcpHandleSse
- API base URL: https://lucidry.base44.app/functions

## Authentication
API key as a Bearer token in the Authorization header, or the X-API-Key header. Create and revoke keys at https://lucidry.io/dashboard/settings/api-keys

## Pricing (per 1,000 words, billed per word)
- Cleanup: $2.40
- Professional Tone: $4.50
- Everyday Language: $6.00
- Write Like Me: $9.00
- Magic Button: $15.00
- Translation: $20.00, multiplied by the language-pair tier (1.0 / 1.2 / 1.5)
- Past 500,000 words a month: a flat $10.00
- Free tier: 5,000 words a month, shared across every skill
- Wallet minimum top-up: $25

## Key Links
- Product: https://lucidry.io/features
- Use cases: https://lucidry.io/use-cases
- Ghostwriters: https://lucidry.io/ghostwriters
- Copywriters and editors: https://lucidry.io/copywriters
- Who it is for: https://lucidry.io/who-its-for
- How Lucidry compares: https://lucidry.io/competitors
- For AI agents: https://lucidry.io/agents
- Agent pricing: https://lucidry.io/agents/pricing
- Translation: https://lucidry.io/translation
- How it works: https://lucidry.io/how-it-works
- Pricing: https://lucidry.io/pricing
- Customer pricing: https://lucidry.io/customer-pricing
- Developer docs: https://lucidry.io/developers
- REST API reference: https://lucidry.io/docs/api
- MCP guide: https://lucidry.io/docs/mcp
- SDK reference: https://lucidry.io/docs/sdk
- Status: https://lucidry.io/status

## Limits
- Max file size: 50MB
- Max pages: 500
- Data retention: 30 days after delivery, configurable down to zero
- Rate limits: per API key tier


## Resources & Tools

- Resources & Tools: https://lucidry.io/help/resources
- How Unverified LLM Output Damages Developer Trust: https://lucidry.io/help/resources/automation-teams/unverified-llm-output-damages-developer-trust, Technical audiences execute documentation instead of just reading it, so unverified LLM output breaks trust instantly through altered identifiers, invented endpoint values, and plausible-looking but w
- The Hidden Revision Tax: What Rework Actually Costs Your Agency: https://lucidry.io/help/resources/agencies-ghostwriters/hidden-revision-tax-what-rework-costs-agency, Every agency working with AI drafts pays a tax it never invoices: the rework loop. The draft got cheap while the revision did not, and the margin evaporates in the gap between fast drafts and finished
- Why Every AI Content Tool Stops at the Draft: https://lucidry.io/help/resources/automation-teams/why-every-ai-content-tool-stops-at-the-draft, Generation is solved. Almost every AI content tool stops exactly at the line where the work becomes valuable, and the human editor becomes the integration layer carrying context between five half-tool
- The Black-Box Rewrite: Ask Vendors to Show the Work: https://lucidry.io/help/resources/publishers-consultants/black-box-rewrite-ask-vendors-show-the-work, A rewrite you cannot see is a rewrite you cannot trust. Content tools hide their work as a business model, because a tool that shows its work can be held to it, while a results-only box is judged by v
- The Polished Error: When Better-Sounding Copy Is Worse: https://lucidry.io/help/resources/agencies-ghostwriters/polished-error-when-better-sounding-copy-is-worse, The most dangerous content failures read best. A polished error improves the sentence while breaking the fact, and no grammar tool or readability score catches it, because those tools reward exactly t
- HTTP 200 Is Not Content Success: Acceptance Criteria for Agent Pipelines: https://lucidry.io/help/resources/automation-teams/http-200-not-content-success-agent-pipelines, In an agentic pipeline, a successful API call is not a successful piece. Status codes measure transport, not truth, so green dashboards happily ship content that says 40 languages when the source says
- From Source to Trusted Artifact: The Publishing Standard: https://lucidry.io/help/resources/publishers-consultants/from-source-to-trusted-artifact-publishing-standard, The unit of content work is not the draft, it is the trusted artifact: traceable to its source, refined under guardrails, validated, and returned with receipts. This piece defines the five properties 
- Voice Fidelity: Protecting the Client's Voice as a Load-Bearing Fact: https://lucidry.io/help/resources/agencies-ghostwriters/voice-fidelity-protecting-clients-voice, A ghostwriter's product is not words, it is the client's voice rendered on the page. Generic tools erase that voice by averaging it toward the internet's default tone, and the client feels it on the f
- Closing the Rework Loop in Automated Content Pipelines: https://lucidry.io/help/resources/automation-teams/closing-rework-loop-automated-content-pipelines, Automated content pipelines lose their margin in the rework loop: the cycles of regenerate, rework, and manual rescue that hide between the success metrics a dashboard actually shows. This piece maps 
- Fact Standards That Survive Automation: https://lucidry.io/help/resources/publishers-consultants/fact-standards-that-survive-automation, Editorial fact-checking was built for human volumes and human authors, and machine-drafted content breaks that safety net in three ways: drift with no fingerprints, fluency that masks accuracy, and cl
- What the AI Changed in Your Content (and What It Did Not): https://lucidry.io/help/resources/agencies-ghostwriters/what-ai-changed-content-refinery, Every AI rewrite touches more than grammar. It can shift voice, drop facts, and flatten the edges that made the piece yours. This field note traces what a refinery actually changes, what it preserves,