MCP code snippets that agents can search, retrieve and apply
MCP gives AI agents a standard way to call tools. Moresq Corpus uses that surface to expose repo memory without forcing long code blocks through model output.
What the MCP layer should do
The MCP layer should be small and sharp: list snippets, search known patterns, retrieve short context, apply approved snippets and return receipts. It should not turn into an unbounded document dump.
Moresq Corpus keeps the agent interface focused on action. Search is cheap, retrieval is budgeted and apply operations are auditable.
search_corpus
Find prior implementations and runbooks by query.
get_snippet
Read the exact snippet and metadata when adaptation is needed.
apply_snippet
Write trusted bytes from the snippet registry into the target repo.
proof receipt
Record source, hash, target path and retrieval context for review.
Why Moresq Corpus is the brand layer
The brand is not a generic comparison to another API marketplace. The brand is a trust layer for agentic code reuse: source-backed memory, governed snippets, retrieval receipts and a learning loop for missed work.