Sean Fraser

Agent infrastructure, security
& operations

Current work

Audit · 2026-09-28

Field notes
Proposed field note

Agents as Code

Next in the working set Next: agent observability
CloudHacks Cloud security archive

01 / Position

Systems thinking for agents in production.

I work where AI agents meet the less glamorous constraints that make them dependable: identity, policy, observability, networking, and operator control.

Built from field work.
Documented without theatre.

02 / Current systems

Working systems, shown with constraints.

Three parts of one operating model: define the agent, observe what it does, and govern what it may reach.

  1. 01

    Define

    Forge

    216 tests passed

    Constraint / Typecheck needs attention

    v0.2.7

  2. 02

    Observe

    Lantern

    Build, tests, typecheck passed

    Constraint / CJS runtime smoke test pending

    v0.3.6

  3. 03

    Govern

    Bastion

    Gateway policy implementation

    Constraint / Clean install dependency blocked

    v0.2.7

03 / Field notes

Operational notes from the archive.

Long-form work on cloud security, infrastructure, and the agent systems now growing around them.

Open the full field-note archive

04 / Imprint

CloudHacks remains the security archive.

The name is retained for the body of cloud and network security writing already published here. Sean Fraser is the operator identity; CloudHacks is the durable imprint.

Browse the archive

05 / Utilities

Small tools for real decisions.

Browser-local calculators for comparing AI model and serverless infrastructure costs without creating an account or sending inputs to a backend.

Inspect the tools

06 / Contact

Working on agent infrastructure that has to survive contact with production?

Available for conversations with technical teams, partners, and hiring leaders.