Repository navigation
Release v7.143.0
Summary
Adds a reverse-engineering skill, a shared evidence format and an understanding pipeline that merges source, running-app, capture and documentation inputs into one Screenplay model.
Added
cratis-screenplay-reverse-engineeringskill: reverse engineers a website, its documentation, network captures or a shipped executable into an evidence-backed Screenplay model scoped to the areas you choose, then renders it and walks the running Stage application to check fidelity against the original. (#1056)- The reverse-engineering fidelity check keeps a ledger of observed behaviors, closes a row only when every observed case kind passes on the running app, requires a negative control for each check, and never counts an unrun or undecided check as a pass. (#1056)
- One starting point for understanding an existing system: point the event-modeling skill at any mix of source code, a running system, captures and documentation, and it routes each input to reverse engineering, legacy extraction or Prologue capture and merges the results into one Screenplay model. (#1066)
- A shared evidence format across reverse engineering, legacy extraction and Prologue, so a rule seen in a capture, stated in documentation and found in code is one row with combined strength. (#1066)
Changed
- The reverse engineering and legacy extraction skills lead with systems you own and migrations; rebuilds from public material alone are labeled plausible. (#1066)
- The reverse engineering skill now probes the full query surface beyond the UI, confirms surprising results with a second probe, tests field combinations and one past each limit, and adds a first-validation checklist, a recipe for reaching a running Stage application and a fallback when the Screenplay MCP server is unavailable. (#1056)