Development Guide
August 12, 2026 ยท View on GitHub
Entry point for VT Code contributor workflows.
Getting Started
- Development Setup - Canonical local setup and quality loop.
- Testing Guide - Test commands, structure, and benchmark coverage.
- C++ Core Guidelines Adoption - Policy for any C/C++ code introduced in this repository.
- CI/CD - Pipeline behavior and verification stages.
- Cross Compilation - Multi-target build workflows.
- Fuzzing -
cargo-fuzzusage and parser hardening.
Security and Execution
- Process Hardening - Runtime hardening controls.
- Execution Policy - Command policy model.
- Command Security Model - Command validation and threat model.
- Security Guide - Process sandbox boundaries, MCP inheritance, and provider diagnostic redaction.
- Runtime Guidance - Boundary between compiled universal behavior and dynamic project instructions.
- vtcode Binary Gotchas - Binary startup, runloop recovery, allocator, and request-assembly invariants.
- AI Tool Surface Migration - Breaking-change notes for the Codex-style default tools.
Performance and Reliability
- Performance Guide - Profiling and optimization workflow.
- Rust Performance Principles - Rust hot-path and I/O guidance.
- Performance Hasher Policy -
rustc_hashusage policy. - Async Performance Audit - Async architecture performance findings.
- Session Event Persistence - Canonical session events, exporter boundaries, retention, and verification.
Model Management
- Adding Models - Complete workflow for adding new LLM models.
- Model Addition Checklist - Step-by-step checklist for model additions.
Maintenance Workflows
- Asset Synchronization - Embedded asset maintenance.
- AI Tool Surface Eval Report - Executable tool-surface cases, suite validation, and the baseline comparison format.
- Changelog Generation -
git-cliff-based changelog updates. - Desire Paths - Known architecture pressure points.
- TUI-Only Refactoring Notes - Historical refactor details.
- Tool Summary Display - Compact and expanded tool transition summaries, configuration, and testing boundaries.
- First-party debt scan - Detect actionable debt markers while excluding generated and fixture content.