# Context7 launches Docs7, a Mintlify-compatible documentation host built for agents > Source: > Published: 2026-09-03 19:52:29+00:00 # Context7 launches Docs7, a Mintlify-compatible documentation host built for agents **Upstash's new product serves Markdown, WebMCP search and agent analytics, with a free tier and a $200 monthly Pro plan.** By [Ryan Merket](/author/ryan-merket) ยท Published Primary source: [Context7 on X](https://x.com/Context7AI/status/2095586624868409756) ## Why it matters Context7 is converting its position inside coding agents into a documentation business, using retrieval data to identify missing material and sell the workflow that fixes it. Upstash founders Enes Akar and Mehmet Dogan launched [Docs7](https://context7.com/docs7) on September 3rd, turning Context7's documentation retrieval service into a hosted publishing platform for software teams trying to make their products legible to AI coding agents. [https://x.com/Context7AI/status/2095586624868409756](https://x.com/Context7AI/status/2095586624868409756) The [announcement from Context7](https://x.com/Context7AI/status/2095586624868409756) pitches Docs7 as a way to publish clean, agent-readable documentation with native WebMCP and Agent Skills support, analytics for AI traffic and controls over the material agents can retrieve. The product still generates a conventional documentation website for human readers. Akar and Dogan built Upstash around a similar simplification strategy. In a [2024 account of Upstash's history](https://upstash.com/blog/story-of-upstash), Akar described the founders as developers frustrated by the latency, configuration and maintenance burden attached to infrastructure products. Upstash initially applied that approach to serverless Redis, followed by messaging, workflow and vector products. Context7 moved the group into retrieval infrastructure for coding assistants. Docs7 moves one step upstream by hosting and maintaining the source material those assistants consume. That shift gives Upstash a direct route from agent demand to documentation production. Context7 already sees the questions coding agents ask, the pages they request and the searches that fail. Docs7 packages that data into a publishing product, then offers an agent that can draft documentation changes from the gaps. Context7 says its service has processed more than 100 million agent requests, including more than 18 million in the last 30 days, across 122,000 indexed libraries. Its launch thread separately claims Docs7 sites can reach more than 100,000 active Context7 developers. Those figures are supplied by Context7 and do not disclose how many people generate the requests or how activity is distributed across libraries. ### One build for people and agents Docs7 connects to a GitHub repository containing `docs.json` or the older `mint.json` configuration. It supports MDX pages, OpenAPI files, snippets, redirects and themes. A [public command-line tool](https://www.npmjs.com/package/%40upstash/docs7) lets developers preview the site locally, while pushes to the production branch trigger hosted deployments. Pro customers receive preview deployments for eligible pull requests. Each published site exposes normal HTML alongside machine-readable formats. The [agent-readiness documentation](https://context7.com/docs/docs7/agent-readiness) lists per-page Markdown, `llms.txt` , `llms-full.txt` , a JSON search endpoint and an Agent Skills discovery index. Agents can also request Markdown from a normal page URL through the HTTP `Accept` header. Docs7 returns suggested pages when an agent requests a nonexistent URL. That is aimed at a common failure mode in which a model invents a plausible documentation path or reuses an address from an older version. A request for `/config.md` , for example, can be redirected toward a likely `/configuration.md` match instead of ending at an empty 404 page. The built-in WebMCP tool provides another retrieval route. Compatible browser agents can call a read-only `search_docs` tool from the open website, while Context7 handles indexing and retrieval behind it. WebMCP remains experimental, so support depends on compatible clients and browsers. ### Analytics with a necessary denominator Docs7 separates human traffic from requests made by tracked AI products. Its [AI Visibility dashboard](https://context7.com/docs/docs7/ai-visibility) currently identifies traffic from OpenAI, Claude, Perplexity, Cursor, OpenCode, Meta and Mistral. It reports requested paths, countries when available, timestamps and common WebMCP searches. Context7 is explicit about what the metric does not prove. An AI request is a single visit from a tracked product, rather than a unique user, conversation, citation or completed answer. Several requests may come from one research task, and the dashboard identifies the provider without identifying the exact model version. High traffic to a page therefore shows retrieval demand, not whether an agent used the material correctly. Human analytics cover visitors, page views, searches, assistant questions and referrers. The product flags searches that produced no results, giving documentation teams a list of missing subjects or vocabulary rather than another aggregate traffic chart. ### Context7 turns unanswered questions into a paid workflow Docs7's Pro plan adds autonomous documentation agents. The marketing page says unanswered Context7 queries can become reviewable pull requests. The [product documentation](https://context7.com/docs/docs7/agent) describes a more controlled workflow: users select up to 10 queued suggestions for an agent run, while scheduled daily or weekly runs begin only when at least three suggestions are waiting. The agent verifies suggestions against the connected repository and limits edits to the configured documentation path. A separate pull request reviewer can check broken links, build errors, grammar and spelling. Reviews and automated changes are disabled by default for new sites. A planned monthly full-site checkup is not yet available. Docs7 is free with hosting, a custom domain, Context7 integration, WebMCP search, agent analytics and an API playground. The Pro plan costs $200 per month and adds the documentation agents, pull request previews, an assistant, team support and SOC 2-backed infrastructure. Mintlify compatibility is the clearest distribution tactic. Docs7 accepts the established `docs.json` format and existing MDX projects, allowing teams to test a migration without rewriting their documentation. Upstash is betting that Context7's retrieval traffic can pull software teams into a paid hosting product, closing the loop between what agents ask and what developers publish.