Governing agent actions
Turn one tool-using agent action into a reviewable decision: role, resource, scope, confirmation, evidence, and an explicit return path.
Explore the courseCourse hero under final review
AI-generated fictional presenter
One method, eight applications
Generate → judge → gate → retry
The anchor course teaches the shared method. Each applied track uses it for a different AI system or product: one that answers, retrieves, generates, takes actions, or supports a decision. Start at the anchor, then choose the application closest to your work.
generate→
judge (as code)→
gate (go / no-go)→
retry · promote · hold · rollback
Why this isn't a free blog post
Proof, not vibes
Three things no marketplace course bundles together.
◆ published research
Research-backed
Every claim traces to published, public research — OWASP LLM & Agentic Top 10, MITRE ATLAS, RAGAS, MT-Bench, DORA, NIST AI 600-1, C2PA — cited and credited, never hand-waved.
▶ runnable
57 deterministic labs
The 58 lessons are paired with 57 Python labs that run offline, stdlib-only, with the same output every time. Not slides — code you run, then point at your own system.
✓ honest
Shipped vs backlog
Every lab metric — a numeric result from an illustrative synthetic example — is labelled. We teach you to measure your own system, never claim that ours “hits X%.”
The eight tracks
The method, applied to your domain
Start with the Harness anchor, then add the application track that matches your work — or take the whole Harness.
CORE · the method
The Judge-Driven Harness
The domain-agnostic spine: discover failures, write judges as code, validate them against a compact labeled set, gate, retry, calibrate, wire into CI.
EVAL · eval
Ship LLM Apps on Evidence
Start from trace logs (OpenTelemetry): inspect retrieved sources and answers with the RAG triad, keep a reference set, then use reproducible CI checks. A Pareto frontier makes the cost–quality trade-off visible instead of hiding it in one score.
AGENT · agentic coding
Agentic Coding Harness
Claude Code in production: AGENTS.md as contract, specialized subagents, hardening & determinism, red-team your own agents, judge-driven review — governance, not a feature tour.
SEC · security
Securing LLM Apps
Turn prompt-injection risks into repeatable adversarial tests, then into detection rules for a SIEM: the system that centralizes logs and raises alerts. OWASP and MITRE ATLAS provide the public risk vocabulary.
DOG · dogfooding
Dogfooding
Validate your own product: run it as the eval, orthogonal quality dimensions, calibrated gates, the living question bank, and dogfooding wired into CI — ship on evidence, not "looks fine."
SIM · synthetic users
Synthetic Users & Personas
Goal-directed synthetic-user simulation (an NPC is a simulated user with a goal), multi-turn exploration, and the honest breadth/mass validity boundary: personas widen the behavior space, but do not replace calibration with real users.
DATA · the data layer
The Data Flywheel
Build an evaluation bank from real failures: cover important cases, remove duplicates, version the set, and retain regressions. κ agreement is the check that independent reviewers label the same case similarly.
COST · the efficiency layer
The Cost & Latency Harness
Make the quality, cost, and latency trade-off visible. Compare latency percentiles rather than only averages, route simple requests economically, and use a Pareto frontier to show which cost–quality options are genuinely different.
Standalone public preview
Start with one bounded agent action
This course is outside the eight-track Harness offer. Its public preview opens the course map and GOV.1; enrollment and checkout are not available yet.
Pre-launch offers · individual self-study
Explore the scope and price of each course offer.
The prices below describe three planned offers: the anchor method, the complete program, and a later anchor-plus-track bundle. Offers are localized in BRL for Brazil and USD internationally. Enrollment remains unavailable until the Hotmart purchase and delivery journey passes preflight.
Core
The anchor method — the spine every track builds on.
US$150
Launch: US$120
- Free primer + 9 anchor modules
- 8 runnable labs
- The full generate→judge→gate→retry method
- 12 months of package updates
Best value
The Full Harness
Everything — the anchor plus all seven tracks.
US$500
Launch: US$400
- Everything in Core — the method every track applies
- Block an unsafe LLM release · ship on evidence with traces + release gates
- Govern a fleet of coding agents · dogfood your product as the eval on every change
- Simulate users to their validity boundary · compound an eval set that gets harder to fool
- Route cheap-escalate-hard and gate cost & latency under budget
- All 58 lessons · 57 runnable labs · every research brief · 12 months of package updates
After initial launch
Complete bundles
The anchor + one track, focused on your domain.
US$300
Planned offer: US$240
- Anchor + one full track (or the Validate pair: dogfooding + synthetic users)
- All that track's labs + briefs; Coding Complete pairs the AI Coding OS pack
- À la carte — security · eval · agentic coding · dogfooding · synthetic users · data · cost
- Free primer included · 12 months of package updates
15-day guaranteeRequest through Hotmart under the rules shown at checkout.
Yours to keepPersonal license for the downloaded release, with 12 months of package updates.
Offline deliveryDownloadable bilingual ZIP in your Hotmart purchase library after payment confirmation.
Focused supportEmail help for access, download, and a reproducible first-run issue.
Study time is not published as a promise yet. We will state a range only after observing beginner sessions, rather than guessing from the author’s pace.
This route shows the international USD offer. The checkout total is authoritative; local taxes, payment-method costs, or conversion into another currency may change the amount shown before confirmation. The Level-0 primer remains free.
The AI Coding OS pack → is a separate companion toolkit with 18 install-ready skills — not a sixth course or a ninth Harness track.
About the offers
Straight answers
Is this beyond a free OWASP page / blog post?
Yes. The free material tells you the risks exist. This teaches you to build the harness that
catches them at scale — red-team as a repeatable suite, detection-as-code, judge-driven gates — with
runnable labs and paper provenance. The differentiator is the method + the code, not a list.
Are the labs real, or toy problems?
Real and runnable: 57 Python labs, offline, stdlib-only, deterministic (same output every
run). Numbers are illustrative by design so you learn the pattern, then swap the single HOOK for your own
model and measure your own system.
Is this current for 2026?
The content tracks the 2026 frontier: OWASP LLM (2025) + Agentic (2026) Top 10, current
OpenTelemetry GenAI conventions, MT-Bench judge-bias findings, NIST AI 600-1, C2PA + SynthID, the EU AI
Act Article 50 timeline. Tool status (e.g. OpenAI's announced agreement to acquire Promptfoo) is flagged with OSS alternatives.
Isn't Claude Code already taught free by Anthropic?
The free courses teach the features — subagents, MCP, skills, hooks. The Agentic
Coding track teaches what they don't: governance, hardening, red-team, and quality gates
— Claude Code in production, by people who ship with it — with runnable judge-labs and a starter you keep.
It sits above the feature tier, not next to it.
Individual only — is there a team version?
Today every tier is individual self-study, priced accordingly. A team tier with progress
tracking and seats is on the roadmap, not shipped — so we don't charge for it yet.
How do I receive it, and does it work offline?
After payment confirmation, the Hotmart library delivers a
.zip with an index.html launcher.
Open it in any browser, read on any device, and run the labs on a desktop. The course itself does not phone home;
package updates remain available in the purchase library for 12 months.Does the purchase include a certificate or live support?
No. This launch is individual self-study: it does not include a certificate, academic credit, live tutoring, statistical consulting, or implementation consulting. Email support covers access, download, and reproducible first-run issues.
Which payment methods and installments are available?
Hotmart may offer Pix, boleto, and card depending on the buyer and location. The payment methods, installments, taxes, conversion, and final total shown in checkout are authoritative.