Home

Last Mile 360

Last Mile 360 is a production-readiness scanner that evaluates codebases across five critical dimensions — security, database, infrastructure, observability, and quality — using AI-powered analysis with Semgrep-style static rules, dependency auditing, and LLM-driven contextual reasoning. Inspired by the Norton 360 trust model, it provides a single 0–100 score and actionable fix suggestions so developers can ship with confidence, not anxiety.

Quick Links

Section	Description
Architecture Overview	Cloudflare-native design, PANOPTES orchestrator, zero origin servers
Norton 360 Trust Model	The 10-point trust model mapping Norton principles to code security
Source Repo Analysis	Complete risk matrix of all 16+ evaluated open-source repos
Agent Design	How the 5 specialized agents work (Security, Database, Infra, Obs, Quality)
Scoring Rubric	Category weights, severity deductions, grade definitions
Semgrep Rules	All 14 SAST rules with CWE mappings
CLI Reference	Commands, flags, output formats, exit codes
Configuration	Full .last-mile.yml reference
Auto Fix Engine	How fix suggestions and patch generation work
Inference Strategy	Three-tier AI model strategy and why no self-hosted models
Cloudflare Services	Every Cloudflare product used and what it replaces
Security Posture	How Last Mile secures itself
Build Roadmap	Phase-by-phase progress from Phase 1 through Phase 6
FAQ	Hard questions answered honestly

Current Status

Phase	Status
Phase 1: Core Scanner	✅ Complete
Phase 2: Agent Expansion	✅ Complete
Phase 3: Configuration	✅ Complete
Phase 4: Auto-Fix Engine	✅ Complete
Phase 5: Advanced Analysis	✅ Complete
Phase 6: Scale	🔧 In Progress

Links

• Source Code: github.com/itallstartedwithaidea/last-mile
• Issues: GitHub Issues
• Projects: GitHub Projects
• Parent Site: googleadsagent.ai