Agent Ready: Definition of Ready Checks for AI Coding Agents

Agent Ready checks every Jira issue before an AI coding agent — Claude, GitHub Copilot, Cursor, Rovo Dev — or a person picks it up: acceptance criteria, steps to reproduce, open blockers, an estimate, the repository, and no passwords, keys or hidden instructions. The same rules every time, with no AI in the checks.

Readiness on every issue

Agent readiness sits in the issue’s right column. Collapsed, a lozenge says Ready or a score such as 5/8; open, each missing check says what to add. Copy missing sections, write a criterion step by step (Given, When, Then), or let the Rovo agent draft them from the issue’s own text — then Insert into description adds only what is missing.

Passwords, API keys and tokens are found and shown only as a mask. Text written to an AI agent — “ignore previous instructions”, invisible characters, hidden comments — is found too, and never reaches the agent.

Your backlog’s Definition of Ready, in one look

Right after the install, Backlog X-ray checks the open issues in every project you can see: how many are ready, what is missing most, and which are one check away — fix those first, in the row. Pick Light, Balanced or Strict for agents for every project in one click, and see the trend week by week.

A gate before the agent

The Issue is agent-ready workflow validator holds a transition such as Ready for agent. An unready issue assigned to an agent can go back to the previous assignee with a comment that names what is missing. agentReady = "true" and agentReadyScore < 80 work in JQL, boards and Automation.

The project page Agent readiness is a refinement list for people too: the open issues that are not ready, what each misses, and the agent-suitable ones an agent can take now.

Coding agents work through it

With Agent Ready’s MCP tools turned on, Claude Code, Cursor and other agents take the next ready issue, start it (an unready one is refused), ask the team a question in Jira instead of guessing, submit a plan for risky areas, and finish with a receipt: what changed, the tests, and which acceptance criterion each test proves. The AI agents report shows what was handed, stopped, done the first time and given back, what it cost, and each agent’s tool use.

Rules that fit your project

Project admins set how each issue type is checked, the rules that differ per type, your own required sections, the agent brief (an AGENTS.md for the project), roles, skills and the repository map. A Preview shows the checks a sample issue would get before you save.

Also included

• Ask Rovo — the Agent Ready agent drafts what is missing, splits a too-big issue into sub-tasks and answers “Is the next sprint ready for agents?”; it asks before every change.
• Agent skills — the company’s and each project’s how-tos for agents; import SKILL.md, a .zip or a .tgz.
• Sprint readiness — this sprint or the next, “7 of 12 ready”, on the readiness page and a dashboard gadget.
• Audit trail — every handoff, waiver, answer, plan decision and receipt, as CSV or JSON.

Runs on Atlassian

Agent Ready is built on Atlassian Forge and makes no calls outside Atlassian. Every read with a person behind it is made as that person, so nobody sees more through Agent Ready than in Jira. See the privacy policy.

Related apps

• JQL Blocks — build Jira searches from blocks, with sprint dates as fields.
• Secret Sentinel — find and redact leaked keys in Jira and Confluence the moment they are saved.
• Team Checklist — a Definition of Done checklist on every issue.