VibeCop onboards in five steps: connect your repository, a no-LLM quick scan, stack confirmation, a Stack Profile knowledge base, and your first full analysis.
Sign in and authenticate with GitHub, then select the repository you want VibeCop to analyze. VibeCop registers a webhook on the repo so future pull requests trigger automatic reviews, and reads your default branch to build an initial Pattern Fingerprint of how your codebase solves recurring problems.
The quick scan runs in about 30 seconds with no LLM involved: dependency CVE checks read your manifest, while secret detection, SAST pattern matching, and infrastructure-as-code checks sample the highest-signal source files in your repository. You see real findings immediately, before any AI-powered architectural analysis begins.
After the quick scan, VibeCop detects your languages, frameworks, and package manager, then asks you to confirm or correct the detected stack. It builds a Stack Profile knowledge base from your confirmed stack, which calibrates VibeCop's LLM-powered detector agents to your specific technology context before deep analysis runs.
The full analysis extracts a Pattern Fingerprint from your main branch, runs all detector agents, and computes your first Architecture Integrity Index — a 0-100 score of your codebase's structural health. Every subsequent pull request is reviewed against this baseline, and drift from it is flagged automatically.