Flagship project

Aegis

The AI QA copilot you point at a repository.

Starting QA on an unfamiliar codebase means days of spelunking before you can write a single meaningful test. Aegis compresses that to one scan. It maps the app's surface area (routes, endpoints, auth), scores every route for risk, and generates the artifacts a QA engineer needs on day one. All from one scan console and one dashboard, running on your machine.

View on GitHub ← Back to portfolio

Designed and built by Greg Blacketter · Portland, OR

One scan in, a QA toolkit out

Diagram: your repository feeds one Aegis scan, which discovers routes and auth, scores risk, generates tests, and writes docs, producing a risk map, test plan with runnable tests, app docs, and prioritized issues and fixes.

You choose the depth and the spend

Diagram: three tiers rising like a ladder: Quick Scan (deterministic, instant), Local AI (on-device Ollama model), Cloud AI (configured cloud model, the deepest pass).

Every report runs at any tier. Start free and instant; go deeper only where it's worth it.

What one scan gives you

Risk Map report: routes listed with security, performance, and reliability severity bars; 7 high, 12 medium, 23 low, with auth and payment surfaces flagged.
Risk Map

Every route scored for security, performance, and reliability. A quick rule-based pass, or a deeper evidence-based one.

Test Plan report: a test pyramid targeting 60 percent unit, 30 percent integration, 10 percent end to end, next to prioritized areas and a sample checkout scenario.
Test Plan

A prioritized strategy: test-pyramid targets, priority areas, and sample end-to-end scenarios.

Terminal running the generated suites: pytest, behave, and playwright all passing, 48 tests across 3 frameworks.
Generated Tests

Runnable Behave, pytest, and Playwright projects. Working test suites, not snippets.

App Docs: a critical user journey graph from login through checkout, plus a feature map noting 14 features and 3 detected integrations.
App Docs

Feature maps, integration detection, critical-user-journey discovery, architecture diagrams, and a repo overview.

Issues and Fixes report: a non-destructive diff of formatting and import cleanups beside five recommendations ranked by risk score.
Issues & Fixes

A non-destructive issue report plus prioritized testing and fix recommendations.

Deterministic analyses grid: accessibility, observability, contracts, and compliance, each with a progress bar and findings.
Deterministic Analyses

Cross-cutting coverage (security scanning, accessibility, observability, contracts, compliance), users & roles, pipeline & gates, and test data & environments.

Local-first: your code never leaves your machine

Aegis never houses your repositories. Scans run on your hardware, and the Local AI tier uses an on-device model, so consulting engagements, client code, and pre-release work stay exactly where they are. Cloud models are opt-in, per report, and you always know which tier ran.

What's next

Multi-user Aegis is in active development. A security-first account hub is built and under test, with invite-only access on the way, so teams can share scan results and published artifacts without ever sharing source code. Ahead of that: task-aware model routing (Aegis picks the right model per report) and a full audit trail of every AI decision.

Pairs with Bugalizer

Aegis works the prevention side: find the risks before they ship. Bugalizer works the intake side, an AI pipeline that takes incoming bug reports and validates, triages, localizes the code, and proposes fixes automatically. Together they cover the quality loop in both directions.

Bugalizer on GitHub

Want a walkthrough?

I'm demoing Aegis to QA folks now. If you'd like to see it run against a real repo, or talk about the problems it solves, send a note.

Email jblacketter@gmail.com LinkedIn gregory-blacketter