Sangkwun Sandy
sangkwun-sandy is a Claude Code plugin for the Ship phase that records MCP scenarios and replays them deterministically without LLM inference.
by Sangkwun · github.com/Sangkwun/sandy
Record MCP tool-call scenarios once and replay them deterministically so agent integrations can be tested without burning tokens on repeated LLM inference.
Add it to Claude Code
Install the plugin in Claude Code. One command, paste-ready.
/plugin install sangkwun-sandy@Sangkwun/sandyBuilt to be called by your agent
Skillselion is itself an MCP server. Your agent can pull this entry and a paste-ready install config straight from the API - no copy-paste.
Retrieve this entry with skillselion.get_details("plugin:Sangkwun/sandy") and the paste-ready config with skillselion.get_install_config("plugin:Sangkwun/sandy").
What it does
sangkwun-sandy is a Claude Code plugin marketplace entry for Sangkwun/sandy that targets deterministic MCP scenario replay for AI agents. Builders who wire agents to MCP servers often re-run the same conversational paths to see if tools fire correctly, which wastes latency, money, and attention when every pass needs fresh LLM inference. Sandy’s positioning is create once, replay without the model: you record a scenario of MCP-related calls and play it back through a Sandy player so behavior stays repeatable. That makes it especially useful for solo developers shipping agent features that depend on external tools, where you need fast regression checks on your integration layer. The listing is community-sourced with one plugin and keywords around record, scenario, player, and calls. Install it when you already have MCP workflows and want a lightweight replay harness inside your agent tooling stack, not when you only need generic unit tests with no protocol context.
Highlights
- Deterministic MCP scenario replay for AI agent workflows
- Create scenarios once, replay tool-call sequences without LLM inference
- Sandy player pattern for recorded calls and scenario playback
- Single-plugin bundle from Sangkwun/sandy community listing
- Cuts flake and cost when validating MCP-heavy agent features
Why builders use it
Re-testing MCP agent flows by re-prompting the model is slow, non-deterministic, and expensive for solo builders on tight budgets.
After install, you can record MCP call scenarios and replay them with Sandy so integration checks run the same way every time without another inference pass.
At a glance
- Type - Plugin in Testing.
- Adoption - 0 installs, 13 stars, 0 votes.
FAQ
Who is sangkwun-sandy for?
It is for developers building AI agents that depend on MCP tool calls and want deterministic replay for testing those integrations.
When should I use sangkwun-sandy?
Use it during ship and hardening when you need to verify the same MCP scenario repeatedly without invoking the LLM each run.
How do I add sangkwun-sandy to my agent?
Add the Sangkwun/sandy plugin from the community marketplace in Claude Code, configure MCP scenarios, record once, then use the player for replay runs.
Comments
Share how you use sangkwun-sandy, gotchas, or tips for other indie builders.
No comments yet - be the first to share how you use it.