Backlog Triage Runner

Runs a JQL query against your Jira backlog and triages what comes back: stale items, missing estimates, and work sitting in the wrong state.

Connects to your Jira site with a read-only basic-auth credential, runs the JQL you give it (a project's backlog, open sprints, a specific reporter — your choice), and triages the returned issues: items unchanged for more than your staleness threshold, items with no estimate where the board uses one, issues stuck in the wrong state for their age, and duplicates by title similarity. The output is a triage list ranked by how much each item is blocking the plan, with the specific JQL filter that reproduces each group. Read-only — it never moves a single issue.

Quick Start

ad init --template backlog-triage-runner my-backlog-triage-runner
cd my-backlog-triage-runner
npm install
npm run dev

Tools

• searchIssues: Run a JQL query against Jira (input fields become query params)
• triageIssues: Triage the returned issues: stale, unestimated, mis-stated, duplicates

Required Secrets

• JIRA_SUBDOMAIN — Your Jira site subdomain (from https://.atlassian.net)
• JIRA_BASIC_AUTH — Pre-encoded basic-auth value for Jira: base64 of your Atlassian user plus API token (Atlassian profile → create API token) (Get one)

Try It

jql: "project = ENG AND status not in (Done, Cancelled) ORDER BY updated DESC"

maxResults: "50"

Architecture

• Topology: edge-single
• Archetype: api
• Model: openai/gpt-4o-mini
• Category: devops-engineering
• Industry: saas

Deployment

ad deploy

Generated by AgentDeploy Blueprint Framework