🤝 Top Collaborators
🤓 Latest Submissions
Incisight: Voice Incident Command
During a cybersecurity incident or service outage, teams often lose the thread between spoken updates, evidence, ownership, and stakeholder communication. Incisight gives an incident commander one workspace to keep that record clear. A React dashboard connects to an AssemblyAI Voice Agent for live conversation. The agent can read the current incident, answer questions from its recorded context, and propose timeline entries or stakeholder updates. Every proposed write remains unsaved until an operator reviews and confirms it. The dashboard shows the incident timeline, evidence, conflicting claims, assigned owners, and update status in real time. Incisight also supports service-health checks and advisory lookup through controlled, clearly labeled evidence sources. Its contradiction detection compares structured claims and highlights disagreements for human review. A stakeholder update can be drafted from the incident record, but publishing requires a separate confirmation and approval step. In this version, publication is recorded inside Incisight rather than sent to email or chat. The FastAPI backend validates inputs, stores the incident record, and keeps the permanent AssemblyAI key server-side while issuing short-lived browser tokens. SQLite supports local development; a Render blueprint provides PostgreSQL deployment. The goal is faster incident coordination with an auditable record and human control over consequential communication.
30 Sep 2026
.png&w=828&q=75)
COBOL Build
Many organizations still depend on COBOL programs for important batch processes. Their business rules can be difficult to understand because documentation is incomplete and experienced maintainers are scarce. Rewriting a program in a newer language is risky: a small difference in rounding, record formatting, or error handling can change a financial result. COBOL Bridge is a developer workflow for making that transition easier to inspect and verify. It starts with a COBOL program and synthetic test inputs. IBM Bob 2.0 explains the program and helps identify its record layouts and business rules. The workflow runs the original program and saves its observed outputs as characterization fixtures. Those outputs become the reference for two rewrites: COBOL to Java, followed by Java to Python. An MCP server gives Bob and other compatible clients tools to inspect the source, run the programs, and compare results. The parity report identifies differences by record and field, including formatting and trailer totals. When a result differs, the developer can trace it back to the original COBOL behavior and correct the new implementation without silently changing the reference output. The prototype focuses on one synthetic batch program so the full workflow can be demonstrated clearly. Developers and QA engineers can inspect the explanation, both rewritten versions, the test cases, and the parity reports. The aim is to turn an uncertain rewrite into a series of changes whose behavior can be checked. One sample does not prove that every COBOL system can be converted automatically, but it shows a repeatable approach to safer modernization.
27 Sep 2026
