kcmartin · Oct 20, 2026 · Oct 19, 2026
diff --git a/cli/authentication.mdx b/cli/authentication.mdx
@@ -1,7 +1,7 @@
 ---
 sidebarTitle: Authentication
 title: CLI Authentication
-og:description: Authenticate the Sprites CLI with your Fly.io account
+description: "Sign the sprite CLI in with your Fly.io account, switch orgs, keep tokens in the keyring, and use sprite auth setup --token for CI like GitHub Actions."
 ---
 
 Sprites uses your Fly.io account for authentication. This guide covers setting up authentication and managing API tokens.

diff --git a/cli/commands.mdx b/cli/commands.mdx
@@ -1,7 +1,7 @@
 ---
 sidebarTitle: Commands
 title: CLI Commands Reference
-og:description: Complete reference for all Sprites CLI commands
+description: "Reference for each sprite CLI command and flag: org, create, use, list, destroy, exec, console, checkpoint, restore, proxy and url, plus exit codes."
 ---
 
 Complete reference for all `sprite` CLI commands.

diff --git a/cli/installation.mdx b/cli/installation.mdx
@@ -1,7 +1,7 @@
 ---
 title: CLI Installation
 sidebarTitle: Installation
-og:description: "Install the Sprites CLI on macOS, Linux, or Windows"
+description: "Install the Sprites CLI on macOS, Linux or Windows via script or download, then run sprite upgrade, set a per-directory .sprite file and add completions."
 ---
 
 import CliDownloads from '/snippets/CliDownloads.mdx';

diff --git a/concepts/checkpoints.mdx b/concepts/checkpoints.mdx
@@ -1,6 +1,6 @@
 ---
 title: Checkpoints
-og:description: "Save a Sprite's filesystem at a point in time and roll back to it later. The deliberate save point that complements automatic persistence"
+description: "Snapshot a Sprite's filesystem with sprite checkpoint create, roll back with sprite restore, browse old versions in /.sprite/checkpoints, and prune them."
 ---
 
 Persistence keeps your filesystem across pauses on its own: you don't ask for it, and it's always there. A checkpoint is the other half. It's a save point you create on purpose, a snapshot of the filesystem at one moment that you can return to later.

diff --git a/concepts/connectors.mdx b/concepts/connectors.mdx
@@ -1,6 +1,6 @@
 ---
 title: Connectors
-og:description: Give Sprites access to external APIs without putting provider credentials inside the Sprite
+description: "Let Sprites call GitHub, OpenRouter or any HTTP API through the Sprites gateway without holding the token, using deny-by-default access policies."
 ---
 
 A Sprite that needs to talk to GitHub, OpenRouter, or another HTTP API runs into the same problem: where does the credential live? Pasting a token into the Sprite's environment works, but every Sprite that needs that token now holds a copy of a long-lived secret. Rotating it means touching every Sprite. Auditing access means reading process environments. Revoking it for one Sprite means revoking it for all of them.

diff --git a/concepts/lifecycle.mdx b/concepts/lifecycle.mdx
@@ -1,6 +1,6 @@
 ---
 title: Lifecycle and Persistence
-og:description: "How a Sprite sleeps, wakes, and keeps your data. The model behind warm and cold states and the filesystem that survives them"
+description: "How a Sprite moves from active to warm to cold, what survives each pause, how NVMe and object storage back its ext4 disk, and its 8 vCPU, 100 GB limits."
 ---
 
 A Sprite is not a server that runs until you turn it off. It runs while something is using it, pauses when nothing is, and wakes when work shows up again. Compute comes and goes. Your filesystem stays.

diff --git a/concepts/services.mdx b/concepts/services.mdx
@@ -1,6 +1,6 @@
 ---
 title: Services
-og:description: "Processes the Sprite runtime starts at boot, restarts when they crash, and can wire to your Sprite's URL"
+description: "Create Sprite services with sprite-env so processes start at boot, restart on crash and wake on requests via --http-port. Covers logs and dependencies."
 ---
 
 A Sprite pauses when nothing is using it, and processes don't reliably survive the trip. A warm wake resumes them where they left off; a cold boot drops them entirely. Anything you started by hand in a shell, a dev server, a database, an agent, is gone after a cold wake unless something brings it back.

diff --git a/index.mdx b/index.mdx
@@ -1,7 +1,7 @@
 ---
 title: Sprites Overview
 sidebarTitle: "Overview"
-og:description: "Persistent, hardware-isolated execution environments for arbitrary code"
+description: "Sprites are persistent, hardware-isolated Linux VMs on Fly.io for AI agents, dev environments and untrusted code, billed by the second and free when idle."
 ---
 
 <Frame>![Floating islands enclosed in bubbles, each containing its own isolated world with a server, mountains, and palm trees](/images/sprites.png)</Frame>

diff --git a/integrations/agent-plugins.mdx b/integrations/agent-plugins.mdx
@@ -1,6 +1,6 @@
 ---
 title: Agent plugins
-og:description: Use Sprites from Codex, Cursor, Antigravity, Grok Build, DeepSeek Harness, OpenCode, Pi, Herdr, and Hermes Agent
+description: "Install Sprites plugins for Codex, Cursor, Antigravity, Grok Build, DeepSeek Harness, OpenCode, Pi, Herdr and Hermes Agent, mostly via hosted MCP."
 ---
 
 Agent plugins let an existing coding agent create and operate Sprites without replacing the agent itself. Unless a section says otherwise, the agent stays on your machine and sends remote work to Sprites through the [hosted MCP server](/integrations/remote-mcp/).

diff --git a/integrations/agent-sdks.mdx b/integrations/agent-sdks.mdx
@@ -1,6 +1,6 @@
 ---
 title: SDKs and frameworks
-og:description: Use Sprites with OpenEnv, Google ADK, OpenAI Agents, OpenRouter, and TanStack AI
+description: "Run agents in Sprites with adapters for Hugging Face OpenEnv, Google ADK, OpenAI Agents SDK, OpenRouter Agent SDK and TanStack AI, with runnable examples."
 ---
 
 These adapters let an application use Sprites as the execution environment for an agent.

diff --git a/integrations/claude-managed-agents.mdx b/integrations/claude-managed-agents.mdx
@@ -1,6 +1,6 @@
 ---
 title: Claude Managed Agents
-og:description: Run Anthropic Claude Managed Agents with tool execution inside your own Sprites
+description: "Run Claude Managed Agents tool calls in a per-session Sprite: set up a self-hosted environment, a poller or webhook worker, Tasks keep-alive and cleanup."
 ---
 
 <img

diff --git a/integrations/cortex-code.mdx b/integrations/cortex-code.mdx
@@ -1,6 +1,6 @@
 ---
 title: Cortex Code
-og:description: Use Sprites from Snowflake Cortex Code for persistent builds, tests, and previews
+description: "Add the Sprites plugin to Snowflake Cortex Code, authorize hosted MCP with OAuth, and run dbt tests, Streamlit previews and long builds in Sprites."
 ---
 
 The [Sprites plugin for Cortex Code](https://github.com/superfly/sprites-cortex-plugin) gives Snowflake Cortex Code persistent, isolated Linux environments for builds, tests, previews, and Snowflake development workflows.

diff --git a/integrations/cursor-cloud-agents.mdx b/integrations/cursor-cloud-agents.mdx
@@ -1,6 +1,6 @@
 ---
 title: Cursor Cloud Agents
-og:description: Run Cursor Cloud Agents on your own Sprites from the Sprites dashboard
+description: "Turn a Sprite into a Cursor self-hosted machine from the Sprites dashboard with a Cursor API key, then start, stop or delete it, with computer use."
 ---
 
 [Cursor Self-Hosted Machines](https://cursor.com/docs/cloud-agent/self-hosted) let Cursor run the agent loop and the model while the agent's tool calls, the commands it runs and the files it edits, execute on a machine you control. The Sprites dashboard turns a Sprite into one of those machines with a form: paste a Cursor API key, name the machine, and it appears under **My Machines** in Cursor. No terminal, no scripts.

diff --git a/integrations/index.mdx b/integrations/index.mdx
@@ -1,7 +1,7 @@
 ---
 title: Integrations
 sidebarTitle: "Overview"
-og:description: Connect agents, SDKs, and developer platforms to Sprites
+description: "Choose how to connect Sprites: agent plugins, SDK adapters like OpenAI Agents and Google ADK, Claude and QM platform setups, or the remote MCP server."
 ---
 
 Sprites can be used through the hosted MCP server, a client plugin, an SDK adapter, or a platform's sandbox backend. Choose the path that matches where your agent already runs.

diff --git a/integrations/platform-integrations.mdx b/integrations/platform-integrations.mdx
@@ -1,6 +1,6 @@
 ---
 title: Platform integrations
-og:description: Connect Claude and QM to Sprites
+description: "Connect Claude to Sprites from Anthropic's Connectors Directory with Fly.io OAuth, or deploy QM by Y Combinator with a Sprites sandbox backend."
 ---
 
 These integrations make Sprites available inside another platform rather than through a standalone SDK in your application.

diff --git a/integrations/remote-mcp.mdx b/integrations/remote-mcp.mdx
@@ -1,6 +1,6 @@
 ---
 title: Remote MCP Server
-og:description: Connect MCP clients to Sprites with the hosted remote MCP server
+description: "Point any MCP client at sprites.dev/mcp and sign in with Fly.io OAuth. Covers the mcp- name prefix limit, available tools, the safety model and fixes."
 ---
 
 Sprites runs a hosted MCP server. Connect an MCP client to it and the client can control Sprites directly. It can create and manage them, run commands, inspect checkpoints, and change network policy through standard MCP tools.

diff --git a/keeping-sprites-running.mdx b/keeping-sprites-running.mdx
@@ -1,6 +1,6 @@
 ---
 title: Keeping a Sprite Running
-og:description: Use the Tasks API to hold a Sprite open while an agent or other long-lived process is doing work
+description: "Keep a Sprite from pausing mid-task with the Tasks API on /.sprite/api.sock: create, refresh and delete holds, plus a heartbeat for long agent runs."
 ---
 
 A Sprite pauses in two stages. It first goes `warm`: compute billing stops, the VM suspends, and the next inbound HTTP request wakes it in 100–500ms with process state preserved. If the Sprite stays idle long enough it transitions to `cold`. In-memory state is dropped and the next wake takes 1–2s.

diff --git a/quickstart.mdx b/quickstart.mdx
@@ -1,6 +1,6 @@
 ---
 title: Quickstart
-og:description: Get up and running with Sprites in 5 minutes
+description: "Install the sprite CLI, sign in with sprite org auth, create a Sprite, run commands with sprite exec, and serve HTTP from its Fly.io URL on port 8080."
 ---
 
 <Frame>![A bespectacled bird reading a quickstart guide while a small creature sleeps in a tiny bed nearby](/images/quickstart.png)</Frame>

diff --git a/scripts/generate-cli-docs/mdx-generator.ts b/scripts/generate-cli-docs/mdx-generator.ts
@@ -30,7 +30,7 @@ export function generateCommandsMdx(
   sections.push(`---
 sidebarTitle: Commands
 title: CLI Commands Reference
-og:description: Complete reference for all Sprites CLI commands
+description: "Reference for each sprite CLI command and flag: org, create, use, list, destroy, exec, console, checkpoint, restore, proxy and url, plus exit codes."
 ---
 
 Complete reference for all \`sprite\` CLI commands.

diff --git a/sprite-maintenance.mdx b/sprite-maintenance.mdx
@@ -1,6 +1,6 @@
 ---
 title: Sprite Maintenance
-og:description: Keeping your Sprites happy and healthy
+description: "Upgrade an older Sprite to Ubuntu 26.04, then move it to the newer sbd storage system in place or by copying to a new Sprite, and roll back if needed."
 ---
 
 Sprites are persistent. Everything you leave is there when you come back, including the software. Long-living Sprites still need regular updates. This guide walks through some of the common tasks for maintaining the software environment in your Sprite.

diff --git a/working-with-sprites.mdx b/working-with-sprites.mdx
@@ -1,6 +1,6 @@
 ---
 title: Working with Sprites
-og:description: "Beyond the basics—sessions, ports, persistence, and everything you need to build real stuff"
+description: "Day-to-day Sprites use: sprite exec vs console, detachable sessions, warm and cold wakes, URL auth, sprite proxy port forwarding, checkpoints and SSHFS."
 ---
 
 <Frame>![An assortment of tools and supplies arranged on a workbench shelf](/images/working-with-sprites.png)</Frame>