mcp copilot-instructions.md

Using smplkit — write against it and operate it

smplkit is a hosted developer platform — feature flags, application config, runtime logging, audit, and scheduled HTTP jobs — with two halves you use together: author in code with the smplkit SDKs (resolve a flag, read a config value, install the dynamic-logging adapter, emit an audit event), and operate through the smplkit MCP server (https://mcp.smplkit.com/api/mcp) — express intent ("turn this flag on in prod for enterprise users," "set the staging DB host," "raise this logger to DEBUG," "stream audit events to Datadog," "POST this endpoint every morning at 7") and the tools translate it to the right API calls. Refer to it as "smplkit," not a single-purpose tool for any one capability.

The author + operate loop

The differentiated move is doing both in one flow: write the SDK call in the code, provision the underlying resource with the tools, and verify it — without leaving the editor. The id in code is the same string you provision (a flag id, config id, logger name, audit resource_type).

"Add a kill-switch for the new checkout." → in code, checkout = client.flags.boolean_flag("checkout-v2", default=False) then if checkout.get(): render_new_checkout(); provision with create_flag(key="checkout-v2", type="boolean", default=false); verify with get_flag(key="checkout-v2").

Writing code with the smplkit SDKs

SDKs exist for Python, TypeScript, Go, Java, C#, and Ruby. One client exposes all four runtime products as sub-namespaces (client.flags, client.config, client.logging, client.audit); the SDK key comes from the SMPLKIT_API_KEY env var (or a ~/.smplkit profile); set environment once on the client (production/staging) to scope resolution — never per call. Representative snippets are Python; the per-language reference (install, client init, idiomatic usage for every language) is at https://docs.smplkit.com/products/sdks/<python|typescript|go|java|csharp|ruby>.