How do I self-host AgentFetch with Docker?

Self-host AgentFetch with Docker by pulling the official image and running it on any host with a public port: docker pull ghcr.io/agentfetch/agentfetch:latest && docker run -d -p 8080:8080 -e AGENTFETCH_LICENSE_KEY=... --name agentfetch ghcr.io/agentfetch/agentfetch:latest. The container exposes both stdio (for direct MCP via docker exec) and a streamable-HTTP MCP endpoint at http://host:8080/mcp/. Configure your MCP clients (Claude Desktop, Cursor, Cline) to point at the HTTP endpoint instead of npx. Required env vars: AGENTFETCH_LICENSE_KEY (Pro+ tier or self-hosted license), AGENTFETCH_GLOBAL_QPS (default 100), AGENTFETCH_PER_HOST_QPS (default 1), AGENTFETCH_CACHE_TTL_SECONDS (default 3600). Optional: AGENTFETCH_REDIS_URL to swap the default in-memory cache for Redis (recommended in production), AGENTFETCH_PROXY_URL for outbound HTTPS proxy, AGENTFETCH_PLAYWRIGHT_URL to chain to a Browserless/Playwright container for JS rendering. The container is ~250MB and uses ~100-300MB RAM at idle, scaling to ~1-2GB under load. For Kubernetes, a sample Helm chart lives at github.com/agentfetch/helm-charts. Self-host makes sense when you have data-residency requirements, need >1M fetches/month (cheaper than hosted), or want to deploy AgentFetch on a private network for internal-only agent use.