Rendix — Infrastructure AI that knows your codebase

Claude Code is brilliant for application developers. Rendix is built for the engineers who manage infrastructure, pipelines, and data.

The difference is not whether AI can generate Terraform — it can. The difference is whether it generates Terraform that matches your naming conventions, your tag schemas, your module patterns — without you having to describe them every time.

Rendix indexes your entire repo and retrieves the right context automatically, every generation.

Why engineers switch to Rendix

Most engineers already use AI for infrastructure work. The problem is not the AI — it is the context. When you ask a general AI tool to generate a Terraform resource, it produces a generic example. You then spend 10 minutes rewriting it to match your naming conventions, adding the right tags, replacing hardcoded values with your variable names, and fixing the module references.

Rendix eliminates that rewrite. It reads your repo first, then generates.

# create an RDS instance for the payments service

Instead of a generic example, you get a resource block that uses var.org, var.environment, your existing aws_db_subnet_group data source, your standard tag keys (Environment, Team, ManagedBy, CostCentre), and your provider version constraints. Because Rendix read your codebase before generating.

What it does

Generate infrastructure code — matched to your codebase

Write a comment. Press Cmd+Shift+I. Watch code stream into your editor that looks like it was written by someone who has read every file in your repo.

Supports: Terraform · OpenTofu · Kubernetes YAML · Helm · Dockerfile · Azure Bicep · AWS CloudFormation · Ansible · GitHub Actions · GitLab CI

Your comment stays — Rendix inserts generated code below your comment so you always have the original description as a reference.

Switch files freely — generation runs in the background. If you switch tabs, Rendix keeps streaming and notifies you with a Go to file button when it finishes.

Completion signal — status bar shows ✨ Rendix ✓ and a 🎉 notification confirms the code was inserted.

Automatic accuracy validation — after every generation

This is the feature no general AI tool has. After every generation, Rendix runs a verification pass against your indexed codebase and tells you exactly what it could confirm and what it could not:

Ask follow-up questions directly in the report panel — get corrected code with a Copy button. No other tool does this.

Debug failing pipelines — with full pipeline context

Select the failure output and press Cmd+Shift+D. Rendix auto-detects your pipeline tool — GitHub Actions, GitLab CI, CircleCI, Jenkins, Buildkite — injects the relevant pipeline context, and streams a plain-English diagnosis:

ROOT CAUSE: The npm install failed because the registry was unreachable
FAILED AT:  Install dependencies (line 23)
FIX:        Add a retry step or mirror the registry in your CI config

Ask follow-up questions in the panel. The full context of the failure stays in every answer.

Coming soon: Terminal listener — Rendix watches your VS Code terminal and prompts you to diagnose failures the moment they happen.

Explain, review, runbook and refactor

Explain (Cmd+Shift+E) — select any infrastructure code you didn't write. Get a plain-English explanation with flagged issues and anti-patterns.

Review (Cmd+Shift+R) — select Terraform code. Get a severity-badged review across SECURITY, COST, RELIABILITY, and CONVENTIONS. Every issue includes the specific fix.

Runbook (Cmd+Shift+P → Rendix: Generate runbook) — type a service name or incident description. Get a complete Markdown runbook with symptoms, triage steps, investigation commands, resolution steps, and post-incident actions — built from your actual infrastructure context.

Refactor (Cmd+Shift+P → Rendix: Refactor) — select code. Get it rewritten to match your conventions — hardcoded values become variables, missing tags get added, deprecated arguments get updated.

SQL generation and dbt support

Open a .sql file, write a comment, press Cmd+Shift+I. Rendix reads your schema files, dbt models, and source definitions and generates SQL using your real table names, column names, and warehouse dialect.

-- monthly revenue by customer, incremental

Generates a complete dbt model with {{ config(materialized='incremental') }}, {{ ref('stg_orders') }}, {% if is_incremental() %} logic, and your actual column names — not invented placeholders.

For dbt engineers:

• Generate models with correct {{ ref() }}, {{ source() }}, incremental strategies, and naming conventions
• Review any dbt model for missing tests, undocumented columns, incorrect ref/source usage
• Auto-generates schema.yml entries with column descriptions and recommended tests

For SQL analysts and data engineers:

• Complex queries with CTEs — never nested subqueries when a CTE is cleaner
• Query optimisation — diagnose slow queries and get a fully rewritten version
• Supports: BigQuery · Snowflake · Redshift · PostgreSQL · Athena · DuckDB · MySQL

Every unverified column gets a -- VERIFY: comment. The accuracy report fires after SQL generation too.

Script and YAML review

Open any bash, Python, Go, or YAML file and press Cmd+Shift+I. Get a syntax check first, then severity-badged issues with exact line numbers, a fully rewritten improved version with syntax highlighting and a Copy button, and a follow-up input for questions.

Supports: Bash · Shell · Python · Go · YAML · Kubernetes manifests · GitHub Actions · GitLab CI

Coming soon

• Cost optimisation — scan your existing IaC for oversized instances, missing lifecycle policies, NAT gateways that could be VPC endpoints, and more
• Diagram to IaC — upload an architecture diagram, get complete infrastructure code matched to your conventions

How Rendix compares to general AI tools

The question engineers ask is: "Why not just use Claude Code or GitHub Copilot?"

Both are excellent general-purpose tools. Neither was built for infrastructure engineers specifically.

Rendix is not a replacement for Claude Code or Copilot. It is what infrastructure and data engineers use when they need AI that actually knows their stack.

Local-first privacy

Your code never leaves your machine. Rendix indexes your repo using a local embedding model running entirely on your device. Only the most relevant snippets are included in each API call — your full codebase never touches any server.

Getting started

1. Install the extension

Search for Rendix in the VS Code Extension Marketplace and click Install.

2. Get an Anthropic API key

Go to console.anthropic.com, sign in, and create an API key. Add $5 of credit — that covers hundreds of generations. Typical usage costs less than $1/month.

3. Open a workspace with infrastructure files

Open a folder containing .tf, .yml, .yaml, or Dockerfile files. Rendix indexes your workspace automatically and shows Ready in the status bar.

4. Generate your first resource

Open any .tf file, write a comment, and press Cmd+Shift+I:

# create an SQS queue for the payments service with a dead letter queue

Your API key is requested on first use and stored securely in your OS keychain.

Keyboard shortcuts

Command Palette (Cmd+Shift+P → search Rendix:): Generate runbook · Refactor to match conventions · Reindex workspace · Set API key

Requirements

• VS Code 1.85 or higher
• An Anthropic API key (get one here) — pay-as-you-go, less than $1/month typical usage
• Your repo open as a folder in VS Code — this is how Rendix learns your conventions

Supported file types

Extension settings

Pricing

Rendix is free to install. You bring your own Anthropic API key — typical usage costs less than $1/month.

Paid plans with usage limits, team features, and shared knowledge base are coming soon.

Privacy and security

• Repo indexed locally — local embedding model, nothing leaves your machine
• Only the most relevant snippets sent per API call
• Full codebase never touches our servers
• Anthropic does not retain API inputs beyond the duration of the call
• Full details: github.com/tundeafod/rendix

Feedback and support

• Issues: github.com/tundeafod/rendix/issues
• Email: support@rendix.dev (coming soon)

Release notes

0.0.2 — March 2026

• Updated positioning — infrastructure AI that knows your codebase
• Sharpened README with direct Claude Code comparison
• Fixed marketplace listing links and removed placeholder URLs
• Demo gif added to listing

0.0.1 — March 2026

• IaC generation with repo-indexed RAG context
• Pipeline failure diagnosis — auto-detects GitHub Actions, GitLab CI, CircleCI, Jenkins, Buildkite
• Automatic accuracy validation after every generation — VERIFIED / NEEDS REVIEW / MISSING CONTEXT
• Conversational accuracy report with copy buttons
• Script review — Bash, Python, Go — severity badges, syntax highlighting, follow-up questions
• YAML review with syntax error detection
• SQL generation with schema context — dbt models, schema.yml, source files
• dbt model review and query optimisation
• Explain (Cmd+Shift+E), Review (Cmd+Shift+R), Runbook, Refactor commands
• Background generation — switch files freely, notified on completion
• Trigger comment preserved after generation
• Local-first RAG indexing
• Status bar index state

Upcoming

• Cost optimisation from your existing codebase
• Diagram to IaC
• Terminal failure detection
• GitHub Actions, GitLab CI, CircleCI API integrations
• Live cloud schema connectivity — BigQuery, Snowflake, Redshift, Athena