Selected Agent Experience work

The through-line.

My work treats Agent Experience as a product discipline spanning research, design, and engineering: helping agents move from a person’s intent towards an outcome that can be checked, recovered, and handed back.

The same skills apply across the five systems below.

Systems covered.

Orientation and discovery
Tools and interfaces
Workflows and state
Delegated action and trust
Evaluation and improvement

Inspectable artefacts

Selected public work.

Agent Experience field guide and public MCP
What it is
A static Astro field guide with an MCP endpoint that searches the guides, returns the rubric and checks public tool catalogues.
What I did
Designed the content model, the reader paths, the shape of the tool catalogue and the surfaces agents read from on this site.
Open Agent-Readiness Rubric
What it is
A web guide and JSON twin for profiling one agent job across six tracks, with evidence classes and stoppers.
What I did
Wrote the rubric’s structure and published it with a machine-readable version.
MCP tool-description test cases
What it is
Copyable cases for reviewing selection, schemas, invalid inputs, result shape and recovery in a public tool catalogue.
What I did
Turned the tool-description guidance into cases that people and agents can check a catalogue against.
Awesome Agent Experience
What it is
A public reading list of papers, specifications, tools and references on agent-facing systems.
What I did
Collected and sorted the public sources so the field has one readable trail through them.
Broader product and engineering portfolio
What it is
A separate public portfolio of work outside this field guide.
What I did
Gives context on my other product and engineering projects.

Want to review one workflow?

Start with the brief: describe the task, the path it takes today and where an agent gets stuck.

Open the workflow review brief