What problem does it solve?
This Skill prevents AI-chat decisions, bug fixes, and backlog items from disappearing into stale conversation history by extracting them into dated repository docs and verifying each claim against real evidence.
Core Features & Use Cases
- Bounded Reconciliation Sweeps: Defines the conversation/date range covered so every sweep has a clear starting point and prior boundary.
- Evidence-Based Auditing: Classifies each extracted item as completed, partial, active, not-active, or unknown using PR, commit, file, and test evidence rather than chat claims.
- Survivor Routing: Sends unresolved work to the real backlog, unrecorded decisions to the decision log or ADR flow, and unrecorded approvals to the approval register.
- Use Case: Revisit last month’s AI conversations, pull out every item that never reached the repo, and produce a durable reconciliation document with citations and next steps.
Quick Start
Use the chat backlog reconciliation skill to sweep the specified chat conversations, extract every unrecorded item, audit each one against repository evidence, and generate the dated reconciliation doc with routed survivors.