dlogs

dlogs provides crucial engineering decisions to your coding agents, preventing them from re-litigating settled constraints. Key features include: * Mining decisions from GitHub PR history * Queryable decision context for coding assistants * Append-only decision records with supersede lineage * Integration with Cursor and Claude Code via MCP * Advisory context on agent-authored pull requests This platform addresses the common problem of coding assistants making changes that conflict with previous team decisions. By extracting rationale from merged PRs and review threads, dlogs transforms buried historical context into a readily searchable database. This ensures that both new engineers and autonomous agents have access to critical design and implementation constraints, allowing them to work more efficiently and avoid redundant efforts. Built for engineering teams aiming to improve the consistency and reliability of code changes, especially those generated by coding assistants. dlogs helps maintain architectural integrity and prevents costly reworks by providing real-time decision context directly within the development environment. It focuses on consequential records, ensuring that only relevant, append-only Decision IDs (DECs) with clear lineage are stored and presented. This streamlines development, reduces review cycles, and keeps projects aligned with established guidelines.