Share

Core Concepts

Your Agent's Questions Need a Recipient and a Timer

By Sable Whitford— October 30, 2026

Feature image for article: Your Agent's Questions Need a Recipient and a Timer

When an AI agent stops mid-task to ask a human something, it often reads as a failure of autonomy. In one study, letting coding agents ask about bug reports stripped of key details won back most of the lost ground, lifting solve rates from about 55% to 69.4%. But the "human" answering was another model, holding the full report and standing by. Real people are in meetings, or asleep. The design work is in the ask: who gets it, what it says, and what happens while it waits.

Core Concepts

Your Agent's Questions Need a Recipient and a Timer

By Sable Whitford— October 30, 2026

When an AI agent stops mid-task to ask a human something, it often reads as a failure of autonomy. In one study, letting coding agents ask about bug reports stripped of key details won back most of the lost ground, lifting solve rates from about 55% to 69.4%. But the "human" answering was another model, holding the full report and standing by. Real people are in meetings, or asleep. The design work is in the ask: who gets it, what it says, and what happens while it waits.

Hands-On Walkthrough

Three Ways an Agent Can See a Page, and What Each One Drops

By Sable Whitford— October 30, 2026

Feature image for article: Three Ways an Agent Can See a Page, and What Each One Drops

Before a web agent clicks anything, somebody has already decided which version of the page it gets. The usual choices are a screenshot, the accessibility tree browsers build for screen readers, or the DOM, the browser's full model of the page. Each one makes some questions easy and others impossible.

One benchmark agent abandoned a purchase because Add to Cart sat just below the screenshot. Some business-software pages, read as DOM, run to half a million tokens. When an agent screws up, what it was shown is the first thing to check.

Hands-On Walkthrough

Three Ways an Agent Can See a Page, and What Each One Drops

By Sable Whitford— October 30, 2026

Before a web agent clicks anything, somebody has already decided which version of the page it gets. The usual choices are a screenshot, the accessibility tree browsers build for screen readers, or the DOM, the browser's full model of the page. Each one makes some questions easy and others impossible.

One benchmark agent abandoned a purchase because Add to Cart sat just below the screenshot. Some business-software pages, read as DOM, run to half a million tokens. When an agent screws up, what it was shown is the first thing to check.

Builder Primary Sources

MCP replaces sessions with explicit handles

Read major changes 1 and 2 to see where state goes once sessions disappear.

MCP cancellation only asks work to stop

Its Timing Considerations section: a cancelled request may already have completed, side effects included.

An IETF draft for safely retrying POSTs

A2A 1.0 for delegating to opaque agents

JSON Schema objects accept extra fields by default

Builder Primary Sources

MCP replaces sessions with explicit handlesRead major changes 1 and 2 to see where state goes once sessions disappear.

MCP cancellation only asks work to stopIts Timing Considerations section: a cancelled request may already have completed, side effects included.

Quick links

An IETF draft for safely retrying POSTs

A2A 1.0 for delegating to opaque agents

JSON Schema objects accept extra fields by default

Terms Worth Unpacking

Some words in a spec look too ordinary to need defining. “Active customer.” “On behalf of.” “Scheduled.” Each one settles a question that someone answered earlier, often years earlier: which twelve months count, whose name goes in the audit log, what happens if a government changes the clocks.

A colleague who meets the word fills the gap from context, or walks over and asks. An agent does neither. It acts on whatever definition sits closest in the data, the token or the code, and the result looks like a correct answer.

The filings, standards and papers below had to write their choices down. They suggest where yours may still be implicit.

Terms Worth Unpacking

Some words in a spec look too ordinary to need defining. “Active customer.” “On behalf of.” “Scheduled.” Each one settles a question that someone answered earlier, often years earlier: which twelve months count, whose name goes in the audit log, what happens if a government changes the clocks.

A colleague who meets the word fills the gap from context, or walks over and asks. An agent does neither. It acts on whatever definition sits closest in the data, the token or the code, and the result looks like a correct answer.

The filings, standards and papers below had to write their choices down. They suggest where yours may still be implicit.

Retail metrics

Two retailers, two meanings of “active customer”

Wayfair counts anyone who bought directly from its sites in the past twelve months, then divides a year of revenue by that count to report $591 per customer. Torrid counts identifiable shoppers, in store or online, recognized mostly through its loyalty program. A query for “active customers” returns whichever rule the data encodes.

Identity standards

“On behalf of” hides two kinds of authority

RFC 8693 separates impersonation from delegation. Under impersonation, the agent cannot be told apart from the user within the token’s rights. Under delegation, the agent keeps its own identity, and an “act” claim names it as the party acting. Only delegation lets a log show that software took the action.

Time standards

“Scheduled” can mean a fixed instant or local time

RFC 9557 lets a timestamp carry a named zone such as [Europe/Paris]. It deliberately leaves out future meetings pinned to local clock time, whose actual instant moves if a government abolishes daylight saving. A string stored correctly today can later contradict itself, and the format can flag that conflict but not resolve it.

Policy definitions

OECD maps where definitions of “agentic” part ways