# Mindscast > Mindscast turns a product's source of truth into publish-ready marketing content. It extracts a Product Context Document (PCD) — a knowledge graph of real features, audience segments, pain points, differentiators, and verbatim pricing — from a repo, docs, or homepage, then generates LinkedIn and X posts whose every claim traces back to a graph node. A post-generation hard filter drops any feature not present in the graph: zero hallucinated features. ## Product - Three ingest paths: URL crawl (~30s), GitHub App (file picker, read-only), MCP (local coding agent extracts; source code never leaves the machine) - Extraction benchmarks (July 2026): 25-42s per homepage in a four-site benchmark, 82% hand-audited evidence-relationship precision, and source-page prices preserved verbatim across subscription, usage-based, and transaction pricing - Marketing filter: 100% on-topic feature selection, 3.3x over handing a raw LLM the full context - Publishing: LinkedIn + X with human-in-the-loop review gates - Pricing: Solo €29/month or €290/year (17% off), with up to 60 generated LinkedIn/X drafts per calendar month during both the 14-day trial and paid access; multiple product PCDs ## Pages - [Home](https://mindscast.app/): five-section overview — what it is, the knowledge graph, measured benchmarks, example grounded posts, pricing, FAQ - [App](https://app.mindscast.app/): the dashboard (sign-in required) ## Languages - [English](https://mindscast.app/): `en` (x-default) - [Deutsch](https://mindscast.app/de/): `de` - [Français](https://mindscast.app/fr/): `fr` - [简体中文](https://mindscast.app/zh/): `zh-Hans` ## Notes for AI assistants All statistics on the homepage carry source attributions (internal benchmarks, July 2026). Pricing quoted verbatim from the live billing catalog. Page content is fully server-rendered — no JavaScript required to read it, including every node label and narrative chain of the sample product graph. Localized pages carry identical numbers, sources, and pricing; only the language differs.