
Claude Code found an app's shared contract in two MCP calls before records ran out
A New Stack report says Claude Code found an application's shared contract in two MCP calls before the records ran out. The case shows agents can surface useful context, but still depend on what a system exposes.
MCP gave the agent a narrow path
The New Stack headline says Claude Code found an app's shared contract in two MCP calls, then the records ran out. In the source framing, MCP can get AI agents into APIs, but it does not decide what they should see.
That distinction matters for agentic development. A model can reach a contract quickly when the endpoint exposes the right surface, yet the same agent can be limited by missing records, inaccessible context, or an incomplete view of the application.
Context is still the bottleneck
The same page lists related items in which agents speed up development while data access slows them down, and in which AI agents stall without infrastructure context under an API tax.
Taken together, the headlines point to a practical limit for agentic workflows: discovery can be fast once a contract is visible, while verification depends on the runtime, data, and services around the code.
Sources: The New Stack
SiTech — AI-powered web development
We build fast, modern websites and bring AI into real business workflows. Have a project or a question? We'd love to help.