Greptile adds self-updating knowledge bases

Greptile is turning code review from a diff-scanning task into a memory problem: the moat is durable, repository-specific context rather than another generic model pass.

• –Cross-file and cross-service documentation can help reviewers trace indirect effects that diff-only tools miss.
• –A running history of bugs gives future reviews a codebase-specific checklist instead of relying only on broad patterns.
• –Real-time updates and manual editing are crucial; stale architectural knowledge could create false confidence.
• –Making the knowledge base available to coding agents extends its value beyond pull-request comments.
• –External context from project-management and observability tools could connect code changes to operational reality.