Replace "Connect to [tool]" openers, em-dash asides, and MCP server boilerplate with brief verb-first copy in plugin and marketplace manifests. Resync the README table that mirrors them, adding rows that had drifted out (teaching, ralph-loop, playwright, github, zoom, clay). Also refresh the stale read-only X marketplace copy and clean em-dashes from the Salesforce server URL help text. Co-authored-by: Cursor <cursoragent@cursor.com>
Cursor SDK
Cursor plugin with a single skill that helps users build on top of the Cursor TypeScript SDK (@cursor/sdk). The skill covers the three invocation patterns (Agent.prompt, Agent.create + agent.send, Agent.resume), the top traps for new integrations, runtime choice (local vs cloud), auth, streaming, MCP, error handling, and ready-to-extend patterns for CI, scheduled jobs, chat, and webhooks.
The skill is short by design — it points at focused reference files only when the user's task clearly falls into one of them.
What it includes
cursor-sdk: design and integration guidance for building with@cursor/sdk, plus reference files for runtime choice, auth, error handling, streaming, MCP, advanced features, and integration patterns.
When to use it
Use whenever the user is integrating, installing, or writing code against the Cursor SDK; mentions Agent.create, Agent.prompt, Agent.resume, agent.send, run.stream, CursorAgentError, or @cursor/sdk; wants to run Cursor agents from a script, CI/CD pipeline, GitHub Action, backend service, bot, or webhook; is choosing between local and cloud runtime; is configuring MCP servers for an SDK agent; or is porting REST /v1/agents calls to the SDK.
The skill is the source of truth for the external @cursor/sdk package and is meant to be loaded eagerly rather than answered from memory.
Reference files
The skill keeps the main SKILL.md short and reads a reference file only when the user's task clearly falls inside it:
| If the user is... | Reference |
|---|---|
| Picking between local and cloud runtime | references/runtime-choice.md |
| Debugging auth (401s, missing key, team vs user keys) | references/auth.md |
Handling errors, retries, rate limits, CursorAgentError, result.status === error |
references/error-handling.md |
| Consuming streams, picking event types, cancelling, or stream vs wait | references/streaming.md |
| Configuring MCP servers (HTTP, stdio, transport, auth injection) | references/mcp.md |
Sub-agents, resume, artifacts, listing/inspecting agents, Agent.messages |
references/advanced.md |
| Building a specific integration (CI review bot, triage, chat, webhook) | references/patterns.md |
License
MIT