AWS's managed MCP server: call any AWS API, run sandboxed Python against your account, and load AWS agent skills on demand — under IAM, CloudWatch and CloudTrail.
Cloud & Infrastructure
Fastly MCP Server
by Fastly
v2.1.4
MIT
fastlycdnedgecomputeobservabilityinfrastructure
Summary
Fastly's official MCP server — manage CDN services, deploy Compute, read real-time metrics and troubleshoot edge configuration through natural language.
Features
- Discovery-based tools: list_commands, describe, execute
- Command allowlisting, configurable from a file or inline
- Dangerous-operation protection and prompt-injection hardening
- Optional PII sanitisation and token encryption
- Automatic pagination and truncation of large responses
- Cached result set that can be re-queried without another API call
- Stdio, HTTP and SSE transports
Installation
Set up this MCP server in your favorite AI agent — copy a ready-made configuration below.
Description
Fastly's official MCP server gives an AI assistant controlled access to the Fastly API, so edge configuration work can be done by asking rather than by clicking through the console or memorising CLI flags.
What it can do
- Service management — list services, domains, backends and versions; update backend configuration.
- Real-time observability — traffic, cache hit ratios and performance metrics pulled live.
- Troubleshooting — walk 5xx errors back to the configuration that causes them.
- Security surface — inspect ACL rules and TLS certificates.
- Compute deployments — ship changes to Fastly Compute services.
The security model is the interesting part
An MCP server with write access to a CDN is a serious thing to hand an agent, and Fastly has built for that rather than around it. Commands are allowlisted rather than open-ended, inputs are validated, and destructive operations sit behind explicit protection. There is prompt-injection hardening, optional PII sanitisation on responses and optional token encryption. The allowlist itself is configurable — from a file or inline — so a team can narrow the agent's reach to exactly the operations it should have and no more.
Large responses are paginated and truncated automatically, and a set of cache tools (fastly_result_read, fastly_result_query, fastly_result_summary, fastly_result_list) lets the agent query a stored result set instead of re-pulling it, which keeps a metrics investigation from consuming a context window.
Shape of the tools
Rather than hundreds of one-per-endpoint tools, the server exposes a small discovery-based set: fastly_list_commands to see what is available, fastly_describe to learn a command's parameters, and fastly_execute to run it. That keeps the tool list short and lets the API surface grow without the schema growing with it.
Installing
Written in Go, distributed as pre-built binaries, installable with go install, buildable from source, or runnable from npm as @fastly/mcp (binary name fastly-mcp). Authenticate with the Fastly CLI first. Runs over stdio by default, with HTTP and SSE transports available.
MIT licensed, Tier 1 supported by Fastly.
Related MCP Servers
Google's official MCP server for driving the gcloud CLI from natural language, with allowlist and denylist controls over which commands an agent may run.
Cloud & Infrastructure
1 views
Google's open-source MCP server for enterprise databases — one binary that fronts AlloyDB, BigQuery, Spanner, Postgres, MySQL, Oracle, MongoDB, Snowflake and more.
Databases & DataCloud & Infrastructure
4 views
Datadog's official managed MCP server — query logs, metrics, traces, incidents and dashboards from an AI agent, inside your existing Datadog permissions.
Cloud & Infrastructure