ADR Archeologist

As engineering teams scale rapidly, the reasoning behind major architectural decisions is often lost due to poor or inconsistent documentation. Over time, this erosion of institutional memory creates significant technical debt, slows developer onboarding, and increases the risk of system-level regressions. ADR Archaeologist addresses this problem by functioning as an automated architectural recovery engine built around the core strengths of IBM Bob, including full repository context awareness, multi-step workflow automation, and deep intent inference. The platform is optimized for predictable production environments through three key design principles. First, it relies on state-based forensics instead of volatile Git histories that are frequently degraded by squashed merges, missing commit rationale, or inconsistent commit quality. ADR Archaeologist treats the active codebase as the primary source of truth, analyzing dependency structures, configurations, deprecated utilities, and legacy code scars such as commented-out execution paths to reconstruct surviving architectural intent. Second, the system enforces SLA-bounded execution through a strict linear four-stage engineering pipeline. Unlike open-ended multi-agent debate systems that introduce unpredictable latency and reasoning loops, this deterministic workflow ensures consistent runtime behavior and enables sub-90-second execution suitable for enterprise environments. Third, ADR Archaeologist prioritizes workflow co-location by integrating directly into existing developer ecosystems instead of relying on detached dashboards or static reports that contribute to tool fatigue. Recovered architectural knowledge is compiled into standard Markdown ADR assets directly within the repository under /docs/adr/ and distributed automatically through a GitHub Pull Request pipeline, ensuring that architectural context becomes a living part of the engineering workflow rather than an external artifact.