Moonlight AI Academy
The agentic engineering academy for teams that ship.
Learn to build agents that hold up in production, and to make coding agents a dependable part of how your team ships. Playbooks, reusable skills, and weekly live calls with engineers who do this on real systems.
$745/month. Cancel anytime. No contracts.
Add retry and idempotency to the payment webhook
Scope, acceptance criteria, and the files that matter.
Plans with repo context loaded
Reads AGENTS.md, the payments module, and existing tests.
Tests and release checks pass
Automated gates run before anyone looks at the diff.
Review, then merge or send back
A human engineer owns the decision.
Inside the Academy
- Engineering playbooks
- Agent skills library
- Weekly live calls
- Eval of the Month
01 / The problem
Your engineers have coding agents.
Your org doesn’t have a practice.
Some engineers ship far more. Some refuse to touch it. Nobody agrees what a good agent-written PR looks like.
Every tutorial stops at the demo
Hello-world agents and vendor launch videos. Nothing about evals, permissions, failure modes, or what happens when a tool call loops at 3am.
Adoption is individual, not engineered
Each engineer keeps a private prompt stash. There are no shared specs, repo context, or review standards, so quality depends on who ran the agent.
Leaders own a risk nobody has defined
You’re accountable for code quality, security, and spend. Nobody has written down what “done” means when an agent wrote the code.
The Academy turns agentic engineering into a team practice: written down, reviewable, and owned by your engineers.
02 / What changes
What a team practice looks like
These are the shifts we put in place for engineering teams. The Academy teaches each one, with the templates to do it.
Specs
From vague prompts to scoped specs
Tasks arrive with boundaries, acceptance criteria, and the files that matter. Agents stop guessing.
Context
From tribal knowledge to repo context
Conventions, architecture notes, and commands live in the repo, so every session starts like a senior colleague.
Review
From rubber stamps to real review
A checklist for agent-written code, tests that gate merges, and a human who signs off.
Evals
From eyeballing outputs to evals in CI
Your LLM features get a test suite. Prompt and model changes ship when the evals say so.
Operations
From surprise invoices to measured systems
Traces, cost, and latency for every agent run. You know what it does, what it costs, and when it fails.
Ownership
From one champion to a team that owns it
Shared workflows, clear roles, and leaders who can explain how AI-assisted work gets done.
03 / What’s inside
Three pillars. One membership.
What you build, how you build it, and how you lead the team doing both.
Pillar 01
Curriculum & playbooks
- Development with Agents. Specs, repo context, task delegation, review, and testing.
- Agentic Software. Agent architecture, tool design, evals, observability, and human approval.
- Leading AI Engineering Teams. Rollout, standards, hiring, and team design.
- Eval of the Month. One real eval setup, explained end to end.
Pillar 02
Skills & templates library
- Context files. AGENTS.md and CLAUDE.md starters and repo setup templates.
- Templates. Specs, task boundaries, and review checklists.
- Eval harnesses. Starters for common LLM features.
- Agent skills. Refactors, test generation, migrations, and incident triage, added monthly.
Pillar 03
Weekly live calls
- Builders track. Hands-on agent workflows in real repositories.
- Agentic product track. Architecture and eval reviews of members’ systems.
- Leaders track. Rollout, standards, and team questions.
- Recorded. Every call, for the members who can’t make it live.
04 / How your team will work
A loop your engineers can trust
- 01
Write the spec
Scope, acceptance criteria, and constraints, from a shared template.
- 02
Agent works with repo context
Conventions, architecture, and commands load automatically. Permissions stay scoped.
- 03
Tests, evals, and review gate it
Nothing merges on vibes. Automated checks first, then a human engineer.
- 04
Save it as a team skill
What worked becomes a reusable workflow, versioned in the repo.
05 / The stack
Vendor-neutral, end to end
We teach the patterns, then show them on the tools you already use. We don’t resell any of them.
Coding agents
Claude Code, Codex, Cursor, and others, and which fits which job.
MCP & tool design
Connecting agents to your systems with clear read and write boundaries.
Evals
Datasets, graders, and LLM judges you can trust and run in CI.
Observability
Tracing, cost, and latency for every agent run.
CI & release gates
Where agent output gets tested, reviewed, and blocked.
Models & routing
Choosing models by task, budget, and latency, and switching safely.
06 / The curriculum
12 modules, from first repo
to running the practice
Every module ends with something you apply to your own codebase that week.
Foundations
01–05- 01The agent-ready repo
Context files, commands, and conventions agents can follow.
- 02Specs and task boundaries
Sizing work an agent can finish and a human can review.
- 03Context engineering
What to load, what to leave out, and why sessions drift.
- 04Reviewing and testing agent output
The checklist, the test gates, and common failure patterns.
- 05Permissions and security
Sandboxes, secrets, and read versus write access.
Agentic software
06–09- 06Agent architecture and tool design
When to use an agent, a workflow, or plain code.
- 07Evals that gate releases
Build the dataset, pick graders, and wire it into CI.
- 08Observability, cost, and latency
Tracing agent runs and setting budgets that hold.
- 09Human approval where it matters
Approval points that don’t turn into rubber stamps.
Leading the practice
10–12- 10Rolling agents out to a team
Pilots, champions, standards, and handling skeptics.
- 11Hiring and team design
What to look for, and how roles shift when agents write code.
- 12Measuring without vanity metrics
Signals that tell you it’s working, and ones that lie.
07 / Our edge
Taught by engineers who still ship.
No. 1
Practitioners, not course creators
We build agentic systems and agent workflows for engineering teams. The Academy is what we use there, written down.
No. 2
Vendor-neutral
No reseller deals and no platform to sell you. We compare tools fairly and say when not to use AI at all.
No. 3
Product, workflow, and team
Most courses cover one tool. We cover what you build, how you build it, and how you lead the people doing it.
08 / Is this for you?
Built for engineering teams
that own production
This is for you if…
- You’re a CTO, VP of Engineering, staff engineer, or engineering manager
- Your team uses coding agents, unevenly
- You’re shipping, or about to ship, LLM features to real users
- You want standards your team can follow, not tips
- You’d rather learn from engineers than from a vendor
This probably isn’t for you if…
- You want a no-code tool rather than an engineering practice
- You’re looking for prompt hacks or AGI predictions
- You don’t plan to change how your team reviews and ships code
- You want someone to build it for you. That’s our services work, not the Academy.
09 / Membership
Join Moonlight AI Academy
Monthly membership
$745/month
Under $175 a week. One seat, full access.
Join the AcademyCancel anytime. No contracts.
Bringing a team? Ask about team pricing
Included in your membership
- 12-module curriculum and three playbooks
- Skills and templates library, updated monthly
- Weekly live calls across three tracks, all recorded
- Eval of the Month breakdowns
- Private member community
10 / FAQ
Questions engineers ask first
Is this tied to one vendor or coding agent?
No. We teach patterns that carry across tools, then show them on Claude Code, Codex, Cursor, and others. We have no reseller relationships.
Is it for individual engineers or whole teams?
Both. Each membership is one seat. If you’re bringing several engineers, talk to us about team pricing.
Do we have to share our code?
No. Everything runs in your own environment. On calls, you decide what to show.
Do we need paid AI subscriptions?
You need access to at least one coding agent or model API. The membership doesn’t include those. We help you pick the tier that fits your workload.
Can I cancel?
Yes. Membership is monthly with no contract. Cancel anytime.
Can you just build it for us instead?
Yes. That’s our consulting work: agentic software, development with agents, and engineering leadership. See our services.
Not sure it fits?
Talk to
an engineer first.
Tell us about your team, how you use coding agents today, and what you’re trying to fix. We’ll tell you honestly whether the Academy is the right fit.