Skills

What a skill is allowed to do — declared, not assumed

Every skill ships as a signed manifest: what it can do, what it cannot, the tools it calls and the paths it may read — checked before it is trusted with a single command.

The catalogue

Skills come from a signed vendor catalogue or your own uploads, and can be enabled or disabled per workspace.

Skills come from a signed vendor catalogue or your own uploads, and are enabled or disabled per workspace. Today's catalogue:

  • Active Directory Investigation
  • Exchange Server Triage
  • File Provenance
  • Identity, DNS and EDR
  • Infrastructure Incident RCA
  • Kubernetes Cluster Triage
  • Linux Host Triage
  • Network and Firewall Triage
  • Storage and Filesystem Triage
  • Virtualization Platform Triage
  • Windows Host Triage
The manifest

Every skill — vendor-supplied or your own — carries the same contract. This is Linux SSH Hardening's:

Linux SSH Hardeningv3.4
Owner
Cybersecurity Engineering
Risk
Low
Access
Read-only

CAN

  • Read the effective sshd configuration
  • Inspect include files
  • Inspect the authentication policy
  • Check listening interfaces
  • Check the host firewall
  • Compare against a baseline

CANNOT

  • Read arbitrary home directories
  • Read application secrets
  • Upload files
  • Modify configuration
  • Establish arbitrary network connections

Tools

sshd · systemctl · ss · grep · cat

Allowed paths

  • /etc/ssh/**
  • /etc/pam.d/**
  • /var/log/auth.log

A skill declares what it can and cannot do, the tools it needs, the paths it may read, and how it is evaluated.

How a skill is approved

A skill moves through the same review, whichever catalogue it came from:

  1. Draft
  2. Static validation
  3. Sandbox
  4. AI red team
  5. Owner review
  6. Security approval
  7. Published

Changes wait for human approval before they run.

Employee-authored skills

Anyone in the organization can write a skill for their own team and put it through the same lifecycle before it is allowed to run.

What a skill is allowed to do — declared, not assumed

Request a demo