Documentation
AIOptimize docs
AIOptimize is a static-first auditor for AI codebases. It reads your repo, maps every LLM call against a continuously-updated knowledge base of provider best practices, and emits concrete diffs with cost and latency impact estimates.
Quickstart
Install the CLI and scan your first repo in under a minute.
# Recommended: isolated install with pipx $ pipx install aioptimize # Or into your project $ pip install aioptimize # Run a scan $ aioptimize scan .
Common output formats
# Human-readable terminal output $ aioptimize scan . # JSON for CI or tooling $ aioptimize scan . --format json > findings.json # Markdown suitable for PR comments $ aioptimize scan . --format markdown > REPORT.md # Fail CI if any warn/error findings $ aioptimize scan . --fail-on-severity warn # Suppress specific rules $ aioptimize scan . --ignore D003
Walk through a full example in the Quickstart guide.
v0.1 detectors
More detectors ship each sprint. Current coverage:
| ID | Severity | Finds |
|---|---|---|
| D001 | warn | Anthropic system prompt without cache_control |
| D002 | warn | Deprecated model version (Anthropic + OpenAI) |
| D003 | info | messages.create without max_tokens |
Project documents
Internal product, engineering, and GTM docs that back the CLI.
01_product_overview
Product overview
Positioning, personas, USP, and competitor landscape.
02_prd
PRD
Requirements, user journeys, and phased scope.
03_backlog
Backlog
Epics, features, and user stories.
04_sprint_plan
Sprint plan
Sprints 1–6 with point estimates.
05_architecture
Architecture
Tech stack, system design, data model.
06_development_plan
Development plan
Coding standards and CI/CD setup.
07_testing_plan
Testing plan
Test levels and critical cases.
08_release_plan
Release plan
Milestones, deployment, rollback.
09_gtm
Go-to-market
Pricing, landing page copy, distribution.
10_roadmap
Roadmap
Feedback loops, future work, kill criteria.