What we ship.
Start free, then tools you buy once and own forever, plus one subscription that respects your inbox. Open methodology, transparent pricing on every page, no sales calls, no upsell traps.
For your AI assistant
The AIS campus: a free front-desk MCP server plus product buildings your assistant connects to in one line. Free tools work instantly; keys unlock per building — or a campus pass opens everything. Paid once, nothing renews.
AIS Gateway
The AIS campus for AI assistants: a free front-desk MCP server that lists every building, plus two product buildings — daily CVE prioritization (ranked by real exploitation) and the Agent Flight Recorder (dead-man alerts + resume for autonomous agents). Free tools answer immediately; every tool has an HTTP twin for scripts and cron.
- • claude mcp add --transport http ais-gateway https://api.aislabs.ai/mcp
- • Claude app: Settings → Connectors → Add custom connector (two minutes)
- • Wallet agents can pay per call over x402 — no key needed
Gateway keys
Per-building keys or the campus pass. A key unlocks every paid tool and HTTP route in the buildings its scope covers — no wallet, no per-call payments. Paste it once; new tools in your buildings are yours automatically.
- • CVE building $19 · Flight Recorder $19 (30 days each)
- • Campus pass — every building: $29 / 30 days · $149 / 12 months
- • Paid once, nothing renews · same key works on the HTTP API
AI agent engineering
Build agents that finish, coordinate, and stay safe in production. Start free.
Agent Foundations Starter Pack
New to building agents? Four runnable beginner patterns — prompt structure, structured output, tool calling, context basics — with tests, in plain Python. The free on-ramp to the paid libraries.
- • Runs offline, zero dependencies, no API key
- • Read it in an afternoon; paste it into your project
- • Companion to the daily Agent Foundations series
Agent Foundations Code Library
Stage 0, the whole season: working, tested implementations of every pattern from the daily Agent Foundations series — prompts that don't wobble, reliable JSON, tool calling done safely, context you control. Pure Python, no API key. The free starter pack is four of these; this is all 24 and growing.
- • Grounding · token budgets · schema-first JSON · tool permissions
- • Runs offline out of the box — tests ship with every pattern
- • Buy once — new episodes' code added free
Loop Engineering Code Library
Stage 1 — one reliable agent. Working, tested implementations of every pattern from the daily Loop Engineering series: loops that finish, verify, and stop. Pure Python, no API key. Grows with the series.
- • Stop contracts · maker/checker · state files · worktrees
- • Runs offline out of the box — tests ship with every pattern
- • Buy once — new episodes’ code added free
Graph Engineering Code Library
Stage 2 — many steps coordinating. Ten tested patterns for agent-graph orchestration in plain Python: routing, fan-out/fan-in, orchestrator-worker, supervisor handoffs, checkpointing, human approval. No frameworks.
- • Routing · fan-out/fan-in · orchestrator-worker · handoffs
- • Checkpoint + resume · idempotency · approval gates · observability
- • Every pattern runs offline with a test that proves it
Agentic Systems Hardening Library
Stage 3 — safe in production. Ten defensive patterns for the failures that hurt: prompt-injection quarantine, tool permission gates, secret hygiene, spend ceilings, kill switch, tamper-evident audit log, exfiltration guards.
- • Injection quarantine · least-privilege tool gates · secret redaction
- • Spend ceilings · kill switch · sandbox · exfiltration guards
- • Strictly defensive · runs offline · tested end to end
The Complete Stack
All three engineering libraries in one download — Loop, Graph, and Hardening. The full path from a single agent loop to a coordinated, hardened system. Save versus buying separately.
- • Loop Engineering + Graph Engineering + Agentic Hardening
- • 33 tested patterns across the whole stack
- • One download · free updates as each library grows
Vulnerability management
Rank what to patch by real-world exploitation, not just severity.
CVE Priority Kit
For security teams: a daily-regeneratable prioritization report ranking the CVEs your team should actually patch this week. Plan B scoring. The frozen sample on the kit page is the latest Patch Tuesday, refreshed monthly.
- • Plan B scoring (KEV · EPSS · CVSS · vendor advisory · patch availability)
- • Frozen daily sample · live CLI regen with your own NVD API key
- • MIT companion CLI (cve-rank) for ad-hoc scoring
The Brief
The full ranked CVE list plus the reasoning — what moved, what is being exploited, what to act on — delivered weekly, in plain English. The founding list is open now; no charge today.
- • Weekly delivery — full ranked list, not just the top 3
- • Reasoning, not just rankings
- • Founding price locked at $34.99/mo
Shell & admin tooling
Tested scripts for the systems work you already do — run them, prove them, keep them.
PowerShell & Windows Admin Script Library
Tested PowerShell scripts for the work Windows admins actually do — health snapshots, service checks, disk thresholds, safe log cleanup, inventory, connectivity sweeps. Runs on 5.1 and 7, no modules to install. Grows with the series.
- • Objects out of every script — pipe, sort, export
- • Pester tests ship with every script — prove it on your machine
- • Destructive scripts preview by default (-WhatIf built in)
Pricing at a glance
Every price is public. Digital downloads are final — each product has a free way to evaluate it first (frozen samples, daily videos, free snippets), so you know exactly what you’re buying.
| Product | Price | Model | What you get |
|---|---|---|---|
| Agent Foundations Starter Pack | Free | Free download | Four beginner patterns with tests: prompt structure, structured output, tool calling, context basics |
| Loop Engineering Code Library | $49 | One-time purchase | Stage 1 — one reliable agent: every pattern from the daily series, implemented + tested; free updates as episodes air |
| Graph Engineering Code Library | $79 | One-time purchase | Stage 2 — many steps coordinating: routing, fan-out/fan-in, orchestrator-worker, checkpoints, approval gates |
| Agentic Systems Hardening Library | $129 | One-time purchase | Stage 3 — safe in production: injection quarantine, tool permission gates, secret hygiene, spend ceilings, kill switch, audit log |
| The Complete Stack | $199 | One-time purchase — bundle | All three engineering libraries in one download — 33 tested patterns, free updates as each library grows |
| PowerShell & Windows Admin Script Library | $49 | One-time purchase | Tested admin scripts with Pester tests: health, services, disks, safe cleanup, execution policy, inventory, connectivity; runs on 5.1 and 7 |
| CVE Priority Kit | $49 | One-time purchase | Python CLI + daily-regeneratable reports + methodology docs + MIT companion CLI |
| The Brief | $34.99/mo | Subscription — founding list open now, no charge today | Full ranked CVE list + the reasoning, delivered weekly in plain English |
The Loop Engineering Code Library’s price steps up as the pattern collection reaches milestones — early buyers keep every future update free. One-time purchases check out here by card or crypto and the files arrive by email; The Brief’s founding list is free to join and locks your launch price.
License
One purchase, one license, yours indefinitely. These terms cover every paid product below.
What you can do
- Use it internally across your team or organization, on as many machines as you need.
- Modify it for your environment and paste the patterns into your own codebase.
- Ship what you build with it, including in commercial work.
- Keep using it indefinitely, and re-download updates from your purchase link.
What you can’t do
- Resell it, repackage it as your own product, or publish the source.
- Strip the attribution and re-publish it as your own work.
- Share your download link publicly — links are per purchase.
The companion CLI cve-rank is MIT-licensed and unrestricted. The CVE Priority Kit carries an additional defensive-use condition.
Support
A person reads every message — there is no ticket queue and no bot.
Purchase questions, download or install trouble, licensing for a larger team, or a bug in any pattern. Reply within two business days.