CodeArchaeologist

PROBLEM Every mature company has a critical system nobody dares to touch. It keeps the business running, but its author is gone, it has no tests, and its business rules are buried in the code. When the board asks "How much will modernization cost, and where should we start?", the answer is usually a weeks-long consulting engagement or an unsupported opinion. SOLUTION CodeArchaeologist answers that question in minutes, with evidence. It takes a legacy repository (Python 3, Flask and SQLite in this version) and produces: 1. A verifiable technical dossier. IBM Bob audits the code in read-only mode and delegates in parallel to subagents for SQL, routes, security and dependencies. Every finding cites a file, line range and snippet, and a Python validator rejects any citation that does not exist literally in the code. 2. An auditable migration recommendation. A deterministic engine walks the call graph, SQL and complexity, and scores every route with a visible formula: value x testability x business data / risk. It returns the first cut, the route not to touch first, and a three-wave roadmap with PERT effort. 3. A board-ready DOCX memo with the decision, risk matrix, recommendation and traceability from every figure back to the code. On the controlled FacturaYa sample, it also runs a first Strangler Fig cut: the legacy route and its modern replacement pass the same characterization tests (6 of 6). TARGET USERS CTOs and architects inheriting critical systems, consultancies that need a defensible assessment before quoting, and boards that must approve modernization budgets. HOW THEY USE IT Users upload a repository as a ZIP, or open the ready-made FacturaYa analysis. They watch Bob work live, then browse the risks (each finding opens at the cited code), the call graph and the migration recommendation. They can ask Bob questions about the code from a side panel and download the memo for the board. In the Modernization Studio, they pick target technologies and Bob builds a step-by-step plan, applied only to an isolated copy after explicit confirmation. WHAT MAKES IT UNIQUE - AI proposes, code decides. Bob discovers and explains; reproducible code calculates risk, migration order and effort. If Bob contradicts the engine or invents a figure, its answer is rejected. - Evidence before narrative. In the public showcase, 12 findings are validated and all 14 citations pass. - Execution honesty. Every result is labeled live or imported (a real recorded Bob session). - Secure by design. The analyzed code is data, never instructions, and uploaded code never runs. HOW IT ADDRESSES THE PROBLEM Instead of weeks of manual review, a team gets in minutes a map of the risks, a justified starting point and an effort range, all traceable to the code. CodeArchaeologist does not replace the modernization team: it gives them the map, the order and the evidence to start on Monday.