AI CompendiumA public chart of the language
Agents & Harness

Ticket

The unit of delegated work, with scope and a definition of done.

Definition

A tracked work item containing the goal, context and acceptance criteria. For agentic workflows, the ticket is the input: it is what a headless agent picks up and acts on.

Ticket quality becomes throughput. Vague tickets produce vague diffs and long review cycles; precise ones can be executed and verified without a conversation.

The ticket becomes the agent’s whole universe: it is the input, the scope boundary and the completion criterion. That makes ticket quality a direct input to output quality — vague tickets produce confident work on the wrong problem, and the failure is invisible until review.

How people use it
Queueing work for overnight agents.
"Ten well-scoped tickets in the backlog, each with acceptance criteria. Let the agents work the queue."
Worked example

The ticket that got 80% of the way

An unattended agent implements a change and stops short of the actual goal.

The ticket described the change but not the outcome, so the agent stopped when the diff looked complete. Adding the observable outcome — what a user should be able to do afterwards, and which test proves it — is usually enough to close that gap.

Sources
  • docsBest practices for using GitHub Copilot to work on tasks ↗GitHub docs · 2026

    Recommends bounded tasks with explicit acceptance criteria, and names what is a bad fit for unattended delegation: ambiguous, security-sensitive, production-critical or deeply domain-specific work.

  • blogEffective harnesses for long-running agents ↗Anthropic Engineering · 2025

    The initializer-plus-incremental-sessions pattern for work that outlives a context window, and the four ways it fails: doing too much, losing context between sessions, declaring victory early, not testing end to end.

Related terms