Luc1 is not a scanner and not a chatbot. It is a memory-backed operating layer for authorized red teams: local models, local tools, specialist lanes, and evidence that survives between stages.
The shape is simple: Scout, Judge, Act, Remember. Scout maps services, HTTP behavior, APIs, files, identities, binaries, and campaign context. Judge compresses evidence and asks stronger reasoning what matters. Act validates inside the operator’s tool environment and Rules of Engagement. Remember carries findings, outputs, decisions, and prior context forward so the operation does not restart after every turn.
It is built for environments where sending assessment data to a cloud service is a non-starter. Models run locally through the operator stack. Tools run locally. Evidence stays in the operator environment. The specialist lanes - web and network, identity and cloud, forensics, reverse engineering, malware analysis, social-engineering simulation, and reporting - reflect how real teams divide work.
The current system has demonstrated an end-to-end chain with local models, local tooling, shared memory, and operator-owned targets. The build remains pre-release while broader lanes, reporting, and deployment surfaces continue to harden.