Moresq Corpus

Private developer memory

A private AI coding knowledge base for the way your team actually builds.

Keep approved repository patterns searchable for coding agents without turning private code into an uncontrolled prompt dump or a public marketplace.

How it works

Context with a chain of custody.

The useful unit is not a copied paragraph. It is a reviewed implementation pattern with a source, a trust level and a reason an agent is allowed to reuse it.

01

Connect

Read-only repository source and commit metadata

02

Sanitize

Remove secrets and retain provenance hashes

03

Review

A human approves patterns before they become retrievable

04

Retrieve

Agents get compact context with a signed proof receipt

Private by default

The project remains the tenant boundary. Imported candidates cannot silently become public content.

Useful before generation

Search implementation patterns, runbooks and prior fixes before an agent writes another version from scratch.

Evidence attached

Receipts show source commit, hashes, review state, trust level and whether tests have actually run.

FAQ

Questions teams ask before connecting a private corpus.

What is a private AI coding knowledge base?

A private AI coding knowledge base stores approved implementation patterns, runbooks and repository decisions for a team. It gives coding agents searchable context without making customer code public or using it to train a foundation model.

How is a private coding knowledge base different from chat history?

Chat history preserves a conversation. A private coding knowledge base preserves reusable engineering decisions with source paths, review state, provenance, hashes and receipts that can be checked later.

Can teams control what an AI coding agent retrieves?

Yes. Moresq keeps imported candidates private, requires explicit review before retrieval, and records project ownership, visibility, trust level and receipt state for every reusable pattern.

Does Moresq train public models on private code?

No. The private corpus flow is designed to keep source code inside the project boundary. It stores sanitized candidates, provenance hashes and governed retrieval evidence rather than silently exporting customer code.