Knowledge Base

Governing agents in the wild.

Incident reports, architecture deep-dives, and comparisons — from operators running autonomous AI agents in production.

Incident Reports

Real failure modes from running 23+ agents. What happened, what would have stopped it.

10 articles
2026-04-185 min

My AI Agent Ran a Dangerous Command: The Post-Mortem Playbook

What to do when an AI agent executes a destructive command in production, why LLM-based guardrails fail, and the controls that actually stop it next time.

2026-04-184 min

AI Agent Deleted Files: How to Prevent Destructive Actions

An operator's playbook for stopping AI agents from deleting files, dropping tables, and wiping directories. Real mechanisms, not prompt pleading.

2026-04-185 min

AI Agent Jailbreaks in Production: What Actually Happens and How to Contain Them

A field guide to AI agent jailbreaks in live systems — attack patterns, why prompt-level defenses fail, and the execution-layer controls that actually hold.

2026-04-185 min

When an AI Agent Leaks Secrets to the Log: Causes and Fixes

AI agents routinely leak API keys, tokens, and PII into logs. Here's why it happens, what fails to stop it, and what actually contains the blast radius.

2026-04-185 min

AI Agent Prompt Injection Protection: What Actually Works in Production

Prompt injection is a control-plane problem, not a prompt problem. Here's what I learned protecting 23 live agents — and what stops actual attacks.

2026-04-184 min

AI Agent Sent the Wrong Email: How to Prevent It From Happening Again

An operator's postmortem on AI agents sending wrong emails, plus the deterministic controls that actually prevent recurrence in production.

2026-04-185 min

When an AI Coding Agent Destroys Your Repo: Anatomy of a Wipeout

What actually happens when an AI coding agent deletes a repo, why guardrails fail, and the execution controls that would have stopped it.

2026-04-185 min

Autonomous Agent Governance Failures: What Actually Breaks in Production

A field report on how autonomous AI agents fail in production — the specific governance gaps, real incidents, and the controls that actually hold up.

2026-04-185 min

Autonomous AI Agent Safety Risks: What Actually Breaks in Production

The real safety risks of autonomous AI agents in production: prompt injection, tool misuse, cost runaways, and why wrappers like guardrails aren't enough.

2026-04-185 min

I Run 23 Autonomous AI Agents. Here Are 7 Times They Almost Nuked Production.

Real failure modes from running 23 agents in live production — prompt injection, runaway loops, credit exhaustion, silent drift. What happened, what would have stopped it.

Governance

Principles and patterns for governing autonomous AI agents in production.

2 articles
Comparisons

Sift vs. LLM guardrails, AWS Bedrock Guardrails, Lakera, homegrown solutions.

1 article
How-To

Practical guides to wiring governance, ACLs, and audit receipts into agent stacks.

1 article
Deep Dives

Architecture deep-dives on deterministic kernels, ed25519 signing, fail-closed enforcement.

2 articles