Skip to content
Post Mortem

For people working in tech

Security for people who build and run things

For builders, vibe-coders and AI-tool users: keys, platforms, agents, checklists and incident playbooks.

Lessons for people who build

4 min

What is a CVE? The public list of known flaws

A CVE is a number given to a security flaw that everyone knows about. It tells you to check whether you need to update.
4 min

Zero-days: a flaw with no fix yet

A zero-day is a flaw that attackers can use before a fix exists. You cannot update your way out, but you can limit the damage.
5 min

Secret keys: the codes that must stay private

Secret keys let software use a paid service as you. If one leaks, someone else can spend your money.
4 min

Blast radius: how far can one mistake spread?

Blast radius is how much damage a single failure or break-in can cause.
4 min

Least privilege: give only the access that is needed

Give people and tools the least access they need, for as short a time as they need it.
5 min

Borrowed code: the parts of your app you did not write

Most apps are mostly other people's code. Each borrowed piece is another way in.
5 min

Prompt injection: when text tricks an AI

AI tools read your instructions and the things they are working on in the same place. Attackers hide instructions inside those things.
4 min

AI agents: when a helper has too much power

An AI agent that can change your live systems is a powerful worker, not a chat window.
5 min

The first 30 minutes when something goes wrong

Stop it, keep the evidence, tell people. A calm order beats heroics.
5 min

Learning from a bad day without blaming anyone

A good review asks what made the failure possible, not who to blame.
5 min

Are you ready? A simple checklist

Ten checks that tell you whether you could handle a security problem tomorrow.

All lessons for this group

Playbooks for when it breaks

Act in the first 5 minutes
CRITICAL

My API key leaked

A leaked key lets a stranger spend your money. Cancel it first, investigate second.
Act in the first 5 minutes
CRITICAL

My AI agent is wrecking my database

Tables vanishing, rows changing, an agent insisting everything is fine. Stop it, take its keys, then restore carefully.
Act in the first 15 minutes
CRITICAL

My database or storage might be open to the public

Close the door first, then work out who walked in.
Act in the first 15 minutes
HIGH

My cloud or AI bill suddenly spiked

Treat an unexplained bill as a stolen key until proven otherwise.
Act in the first 5 minutes
HIGH

My AI agent did something I did not ask for

Stop it, take its keys away, then read what it actually did.
Act in the first 30 minutes
HIGH

A package or extension I installed looks malicious

Remove it, then replace every secret it could have read.
Act in the first 15 minutes
HIGH

My site is under attack or behaving strangely

Slow the attack, protect what matters, and keep the logs.
Pause before you act
WATCH

I got an urgent request to approve or pay something

Urgency is the tactic. Verify before you approve.
Reply within the hour
WATCH

A customer says something looks wrong with their account

Take it seriously, check quickly, and reply with facts.
Decide in 15 minutes
HIGH

A change broke something. Should we roll back?

Compare the cost of the bad change with the cost of undoing it.
Within hours of confirming it
WATCH

I need to tell customers what happened

Say what you know, what you do not, and what people should do.
Within a few days
INFO

It is over. Now we need to learn from it

Blameless, factual, and ending in a few real changes.

Practice scenarios

Real cases

All recent incidents for this group

Latest news

Security feed for tech teams

See all

Flaws, bad packages, platform alerts and AI incidents. What happened and what to fix.

Looking for the other side? For non-tech savvy folks