// Product
Dev-AID
Development AI Driver
One setup, nothing new to learn.
Give your AI tools the expertise they're missing.
// The Problem
AI Coding Tools Are Powerful. But They're Flying Blind.
You're in flow, solving a real problem — then a merge conflict pulls you out. An hour later, you're writing boilerplate tests instead of shipping. Your AI assistant generates code, but it doesn't know your codebase, your patterns, or your security requirements. It gives you generic answers to specific problems.
Most developers choose one AI tool and hope for the best. They accept output without verification, skip security checks under deadline pressure, and lose context every time they close a session. This isn't productive development. It's guesswork with better autocomplete.
// The Solution
An Enhancement Layer That Makes Every AI Tool Smarter
Dev-AID isn't another AI coding tool. It's a configuration layer that works inside the tools you already use — Claude Code, Gemini CLI, Cursor, Windsurf, Cline, and Codex CLI natively, plus VS Code, Zed, and JetBrains via MCP. It gives your AI context-aware expertise that auto-loads based on your work, dual-AI review that catches what a single model misses, and security automation on every commit. No new CLI. No context switching. Just smarter development, right where you code.
"Vibe coding" means surrendering control — accepting AI output without understanding it. Dev-AID is the opposite: expert skills that guide AI behavior, verification gates that demand proof, and dual-AI review that catches what single-model "vibe coding" misses.
Discipline over vibes.
Structure over chaos. Evidence over faith.
Backed by Real Results
Based on 100+ case studies across teams of all sizes. Not theoretical gains — measured improvements from real development workflows.
Based on GitHub/Accenture, AT&T, Microsoft, and METR research studies.
// Capabilities
What Dev-AID Brings to Your Workflow
86 Skills: 73 Expert + 8 Process + 5 Core
73 domain skills auto-load based on your stack. 8 process skills enforce TDD and verification gates. 5 core skills run security checks on every commit. Zero configuration required.
Dual-AI Code Review
One model generates code, another reviews it for bugs and security holes. You pick which models — and in which order. Two perspectives catch what one misses.
Local Semantic Search
Find any function or pattern in your codebase in 0.15 seconds. 100% private — your code never leaves your machine. No API costs. Works offline. Forever free.
Security on Every Commit
Automated CVE scanning, SAST analysis, and secrets detection on every commit. Gitleaks, Trivy, and Opengrep run in 10 seconds. No manual steps. No forgotten scans.
Multi-AI Routing
Use one model or many — Dev-AID routes each task to the right one automatically. Send large context to a cheaper model, keep critical code generation on your best one. One provider works. Four cost less.
Intelligent Automation
AI resolves GitHub issues, merges conflicts intelligently, and generates test suites. What used to take 30 minutes now takes under 5. Your time goes back to building.
Verification Gates
Process skills enforce TDD cycles, demand test evidence before marking tasks complete, and follow systematic debugging protocols. No more false completions.
Session Persistence
Auto-saves your progress when a session ends and restores full context on restart. Tasks, decisions, branch state — everything survives. Pick up exactly where you left off, across any provider.
// Why Dev-AID
What Makes It Different
Provider Agnostic
Works natively inside Claude Code, Gemini CLI, Cursor, Windsurf, Cline, and Codex CLI. Connects to VS Code, Zed, and JetBrains via MCP. No vendor lock-in. Switch anytime.
Architect Mode
One AI drafts the plan, you approve it, then another implements. No more throwing away hours of generated code because the approach was wrong from the start.
Privacy-First Search
Codebase indexing and search run entirely on your machine — no tokens sent to any API. Zero cost, zero latency, and your code structure stays private even when using cloud providers.
Git Worktree Isolation
Develop multiple features in parallel. Scope declarations prevent overlap, pre-merge conflict detection catches problems early.
Shared Project Memory
Architecture decisions and code patterns survive across sessions. Team-shared via git — new devs get full context on clone.
Language-Aware Security Scanning
Auto-detects your languages and runs the right SAST tools and dependency audits. Supports 10+ languages out of the box — no configuration needed.
Autonomous Agents
Eight AI agents handle entire workflows on their own — reviewing PRs, generating tests, hunting tech debt, fixing CI. Run them solo or combine into teams. You define the goal. They do the work.
Skill Validators
Every skill ships with compliance checks that run automatically. Catches convention violations and anti-patterns before they reach your codebase. Extensible — drop a validator into any skill.
Free & Open Source
MIT licensed. Competitors charge $10–60/month for less.
- • No subscriptions
- • No usage limits
- • No data collection
Run Any Local LLM
Go fully offline with local model inference powered by Ollama. A setup wizard detects your hardware and recommends the right model across five tiers — lightweight for 8GB laptops, mid-range for daily use, the sweet spot for most teams, deep reasoning for complex tasks, and best-in-class for maximum quality. Zero API costs. Zero internet required.
Complete privacy for sensitive codebases.
// Agent Framework
AI Agents That Do the Work
Stop babysitting your AI. Dev-AID ships eight autonomous agents backed by 16 specialized tools across 4 providers. They read your code, make decisions, execute multi-step workflows, and report back when they're done. Run them solo or combine them into teams with parallel, sequential, or DAG orchestration. You define the goal. They handle the rest.
PR Reviewer
Three expert perspectives in one command — security, architecture, and DevSecOps. Read-only and safe to run on any PR. Catches what manual reviews miss — in seconds, not hours.
Test Generator
Finds untested code, writes meaningful test cases, runs them, and iterates until they pass. Supports pytest, Jest, and Vitest out of the box.
Tech Debt Hunter
Scans for code smells, outdated dependencies, and architectural drift — then prioritizes by severity. Turns someday cleanups into actionable, ranked tasks.
CI Fixer
Reads CI logs, finds the root cause, applies a fix, and verifies the build passes — in one command. No more digging through log files at 2 AM.
Conflict Resolver
Analyzes what both branches intended — not just text diffs. Merges intelligently, then verifies the result compiles and tests pass.
Research Agent
Structured technical research with citations — not hallucinations. Investigates APIs, compares libraries, evaluates tradeoffs, and delivers documented recommendations.
Onboarding Agent
Gets new team members productive on day one. Maps your codebase, explains architecture decisions, and generates a personalized getting-started guide.
Doc Auditor
Catches documentation drift before it misleads anyone. Finds broken links, missing docs for public interfaces, naming violations, and stale references to deleted code — read-only and safe.
Multi-Agent Teams
Combine agents into teams that run in parallel, sequentially, or as dependency graphs. Three reviewers examine the same PR from different angles simultaneously. An architect plans, an implementer codes, a reviewer validates — as a single pipeline. Four pre-built teams ship out of the box. Or define your own.
Every agent runs sandboxed with three safety tiers. Use --dry-run to preview without changes. Built-in budget tracking caps costs. Mix providers within a single team — your architect on one model, your implementer on another.
// How It Works
No New Tools. Just Better Ones.
Dev-AID is a configuration layer. It enhances your existing AI tools through standard config mechanisms — CLAUDE.md, .claude/skills/, GEMINI.md. Your workflow doesn't change. Your tools just get dramatically smarter.
1. Install Once
Copy Dev-AID to your project and run the init script. Five minutes. One time. Works with any project, any language, any framework. No dependencies to install, no accounts to create.
2. Everything Auto-Loads
Open your editor. Dev-AID detects your tech stack, loads relevant expert skills, activates security hooks, and makes agents and teams available from the CLI. You don't configure anything.
3. Ship Faster
Write code with AI that knows your patterns. Review with dual-model verification. Commit with automated security scans. Let agents handle the rest. Push with confidence.
Works Where You Work
Dev-AID integrates natively with the tools you already use. Same capabilities regardless of provider. Switch anytime.
Ready to Ship Faster?
Dev-AID is free, open source, and ready to use. Five minutes to install. Zero new tools to learn. Same tools. Radically better output.