[bot] Instrument CopilotKit (322,738 weekly downloads)

[braintrust-bot]

Summary

CopilotKit publishes an official TypeScript agent runtime, @copilotkit/runtime on npm (with @copilotkit/react-core providing the client-side execution hooks), for building in-app AI copilots and agents. CopilotRuntime owns an agent registry, wires an AgentRunner, and executes agents (including LangGraph, CrewAI, and Mastra-backed agents) in response to runAgent() calls. This repository has zero instrumentation for any CopilotKit surface — no wrapper, no diagnostics channel, no plugin, no auto-instrumentation config, no e2e scenario.

What instrumentation is missing

Execution surfaces exposed by CopilotKit that are not instrumented:

SDK Surface	Description
CopilotRuntime + InMemoryAgentRunner (@copilotkit/runtime)	Server-side agent lifecycle management, request routing, and execution against registered agents
copilotkit.runAgent({ agent }) (@copilotkit/react-core, via useCopilotKit)	Triggers agent execution programmatically
agent.runAgent() (on an AbstractAgent obtained via useAgent)	Direct agent-run execution with message/state management
LLM adapter execution inside CopilotRuntime (e.g. BasicAgent({ model, prompt }))	Model-call execution path for the default in-runtime agent

CopilotKit's runtime is a standalone orchestrator (AG-UI protocol bridge) and does not depend on openai or any already-instrumented package directly — it can back onto LangGraph/CrewAI/Mastra agents or raw model calls, but the agent-run span itself (lifecycle, state sync, streaming events) has no equivalent in existing instrumentation.

No coverage in any instrumentation layer:

• No wrapper function (e.g. wrapCopilotKit()) in js/src/wrappers/
• No channel definitions or plugin in js/src/instrumentation/plugins/
• No auto-instrumentation config in js/src/auto-instrumentations/configs/
• No e2e test scenario under e2e/scenarios/
• Repo-wide grep for copilotkit (case-insensitive) returns zero matches anywhere in the repo

Weekly downloads

Weekly downloads: 322,738 (as of 2026-08-10; window 2026-08-03 to 2026-08-09; https://api.npmjs.org/downloads/point/last-week/@copilotkit/runtime)

Braintrust docs status

not_found. CopilotKit does not appear anywhere on the Braintrust integrations index — https://www.braintrust.dev/docs/integrations — which lists agent frameworks (Mastra, LangGraph, Google ADK, Strands, etc.) with no entry for CopilotKit.

Upstream references

• npm package: https://www.npmjs.com/package/@copilotkit/runtime
• Official docs: https://docs.copilotkit.ai
• CopilotRuntime reference: https://docs.showcase.copilotkit.ai/backend/copilot-runtime
• Official SDK GitHub: https://github.com/CopilotKit/CopilotKit

Braintrust docs sources checked

• https://www.braintrust.dev/docs/integrations

Local files inspected

• js/src/wrappers/*.ts — full directory listing — no copilotkit.ts wrapper
• js/src/instrumentation/plugins/*.ts — full directory listing — no CopilotKit channels or plugin file
• js/src/auto-instrumentations/configs/*.ts — full directory listing (33 files) — no CopilotKit config entry
• e2e/scenarios/*copilot* — no matching scenario directories
• Repo-wide grep for fal-ai|copilotkit|semantic-kernel — no real matches for copilotkit