GAN-Powered CarPlay Issue Analyzer

What This Workflow Does CarPlay protocol stack bugs are notoriously hard to diagnose — the failure chain spans Bluetooth pairing, iAP2 control messages, Wi-Fi association, RTSP session setup, and audio/video streaming, often across millions of lines of logs. Traditional log analysis relies on engineer intuition, is slow, and misses protocol-layer interactions. This workflow automates the entire investigation pipeline: 1. Downloads JIRA logs automatically — provide a ticket number, it fetches everything 2. Reads your source code first — derives log TAGs from the actual codebase, never trusts user-provided keywords blindly 3. Validates the problem is real — confirms the reported time window matches log evidence before committing to analysis 4. Generates a structured root-cause report — with log evidence (file:line), code evidence (file:line), and spec references side by side 5. Iterates until confident — a Critic agent independently re-verifies every claim, scores the report, and forces the Analyzer to strengthen weak points until the score passes a threshold (default 7.5/10) The GAN Architecture Inspired by Generative Adversarial Networks, two specialized agents run in opposition: - Analyzer (Generator) produces analysis reports, trying to build a complete, evidence-backed root-cause narrative - Critic (Discriminator) independently re-searches the same logs and code, verifies every claim in the report, and scores 4 dimensions: log evidence quality, protocol spec accuracy, root cause depth, and actionability The loop runs until the Critic's score passes the threshold, a score plateau is detected, or the max iteration count is reached. This adversarial dynamic catches the failure modes that kill single-pass LLM analysis: hallucinated file paths, wrong message IDs, root causes with no code backing. What Makes It Different Real spec knowledge, not hallucinations. The workflow ships with a curated knowledge base built directly from Apple's CarPlay Addendum R10 specification — complete iAP2 message ID tables (authentication, identification, CarPlay connection, HID, media, Bluetooth, Wi-Fi, power, vehicle status), full wireless CarPlay connection flow, RTSP stream setup, resource management semantics, and pairing protocol details. The Analyzer and Critic both load this before they touch a single log line. Who This Is For - Automotive software engineers debugging wireless CarPlay connectivity, audio/video streaming, or session establishment failures - QA teams that need reproducible, evidence-backed bug reports instead of "it stopped working" - Engineering leads who want consistent, high-quality root-cause documentation across the team