The agentic web

Make the agentic web a scoreable event.

One website at a time, make it possible to examine whether agents can use the web on our behalf.

A website can be easy for a person to read and difficult for an agent to use. Siteline makes that difference something we can examine. Understandable boundaries and legible knowledge help make the web usable by both.

The shared question is practical: can an agent understand the information, recognize the limits, and take a useful next step? These projects address different parts of that question.

The connections

How the parts fit together.

Measure usability

Examine a website through the tasks agents need to perform.

Explain the limits

Make boundaries, refusals and next steps understandable.

Make knowledge legible

Structure knowledge for people and machines.

Graceful Boundaries enables the operational-boundary work in Siteline. Knowledge-as-Code supplies a complementary approach to human and machine legibility.

Explore the projects

Start with the part that matters to you.

Enabling standard

Graceful Boundaries

Discoverable limits, understandable refusals and useful next steps for humans and agents. An enabling standard for Siteline.

Explore Graceful Boundaries

Supporting infrastructure

Knowledge-as-Code

Open infrastructure for simple human and machine legibility of structured knowledge.

Explore Knowledge-as-Code

Reading the evidence

A website score is evidence about the tested scope at the time of the scan. Inspect the report and method before treating it as evidence about a particular task.

Research & resources

Explore the ideas behind the work.

Browse all papers, briefs and available media

These are dated research and design documents. Read them alongside the current project descriptions.

Help move this work forward

Bring your part to the collaboration.

Bring a website and a real task an agent should be able to complete. Help define success, examine the result, and identify what makes the next attempt better.