Setup & Usage Guide

AutoPrioritizer AI for Jira Cloud

Prerequisites

Before installing AutoPrioritizer AI, make sure you have:

1. A Jira Cloud instance -- The App runs on Atlassian Forge and is only compatible with Jira Cloud (not Jira Server or Data Center).
2. Jira administrator access -- You need admin permissions on your Jira instance to install Marketplace apps.
3. An Anthropic API key -- The App uses Anthropic's Claude AI for analysis. You will need your own API key from console.anthropic.com.
4. Atlassian Forge CLI (for API key setup) -- The Forge CLI is required for one configuration step. Install it with npm install -g @forge/cli if you do not already have it.

Step 1: Install from Atlassian Marketplace

1. Go to the Atlassian Marketplace and search for AutoPrioritizer AI.
2. Click Get app and select your Jira Cloud instance.
3. Confirm the installation and accept the required permissions:
   • read:jira-work -- Read issue data for AI analysis.
   • write:jira-work -- Update the priority field on issues.
   • storage:app -- Store your configuration and usage data.
4. Complete the subscription purchase through the Marketplace.

After installation, the App will appear in two places:

• Issue panel -- On every Jira issue, in the right sidebar or bottom panel.
• Project page -- In your project's left navigation under "AutoPrioritizer AI".

Step 2: Set Your Anthropic API Key

This is the most important setup step. The App cannot function without an Anthropic API key.

Why is an API key needed?

AutoPrioritizer AI sends your ticket data to Anthropic's Claude AI for analysis. Anthropic requires an API key to authenticate these requests. You provide your own key so that you have full control over your AI usage and costs.

How to get an API key

1. Go to console.anthropic.com.
2. Create an account or sign in.
3. Navigate to API Keys in the dashboard.
4. Click Create Key and copy the key. It will look something like: sk-ant-api03-...
5. Add a payment method to your Anthropic account (the App uses Claude Haiku, which is very inexpensive -- typically well under $1/month for most teams).

How to set the key in the App

The API key is set using the Atlassian Forge CLI. This stores the key as an encrypted environment variable within Atlassian's infrastructure -- it is never exposed to end users or sent to the browser.

Open a terminal and run:

# If you have not already, install the Forge CLI
npm install -g @forge/cli

# Log in to your Atlassian account
forge login

# Set the API key as an encrypted environment variable
forge variables set --encrypt ANTHROPIC_API_KEY sk-ant-api03-your-key-here

Replace sk-ant-api03-your-key-here with your actual API key.

Verifying the key works

After setting the key, open any Jira issue in a project and look for the AutoPrioritizer AI panel. Click Prioritize. If the key is set correctly, you will see an AI recommendation within a few seconds. If you see an error about the API key not being set, double-check the steps above.

Step 3: Configure Your First Project Profile

Each Jira project gets its own prioritization profile. The profile tells the AI how your team thinks about priority.

1. Open the AutoPrioritizer AI project page (find it in the project's left navigation).
2. Click the Settings or Configure tab.

Criteria Weights

The App uses six weighted criteria to drive prioritization decisions. Set each criterion from 0% to 100% based on how important it is to your team:

Criterion	What It Measures	Example
Business Impact	Revenue, number of users affected, strategic importance	A billing bug affects all paying customers -- high business impact
Customer-Facing	Whether the issue is visible to end users and affects their experience directly	A UI glitch on the homepage is customer-facing; a backend refactor is not
Security & Compliance	Vulnerabilities, regulatory requirements, data protection	An authentication bypass is critical security; a password policy update is moderate
Technical Debt	Code quality, maintainability, scalability	An outdated dependency with known CVEs is high; renaming a variable is low
Effort (Quick Wins)	Favors low-effort, high-value items	A one-line config fix that unblocks a feature is a quick win
Time Sensitivity	Deadlines, SLAs, time-dependent value	A conference demo due next week is highly time-sensitive

Team Context

Write a short description of your team, product, and current priorities in plain English. The AI uses this as background information when making decisions.

We are the Payments team at an e-commerce company. Our top priority this quarter is PCI compliance certification. We process 50,000 transactions per day, so any payment-related bug is extremely high priority. We have a small team (4 engineers), so quick wins are valuable.

Priority Level Definitions

The App automatically reads your Jira instance's configured priority levels (e.g., Highest, High, Medium, Low, Lowest). For each level, you can provide a description of what that level means for your team.

• Highest: Production is broken or a security vulnerability is actively being exploited. Drop everything.
• High: Significant impact on users or revenue. Should be addressed this sprint.
• Medium: Important but not urgent. Plan for the next sprint.
• Low: Nice to have. Address when there is capacity.
• Lowest: Cosmetic or minor improvement. Backlog indefinitely.

If you use custom priority levels, the App will detect them automatically. Just provide definitions so the AI knows how to use them.

Custom Rules

Rules are override conditions that take precedence over the AI's criteria-based analysis. They are checked first and always win.

Each rule has:

• A condition (plain language description of when the rule applies).
• An action (the priority level to assign).

Examples:

• Condition: label contains "security-critical" → Priority: Highest
• Condition: issue type is Bug and component is payments → Priority: High
• Condition: label contains "tech-debt" and label contains "low-risk" → Priority: Low

Rules are matched against the ticket's labels, components, issue type, summary, and description. You can enable or disable individual rules without deleting them.

Using Single-Issue Prioritization

1. Open any Jira issue.
2. Find the AutoPrioritizer AI panel (usually in the right sidebar or at the bottom of the issue view).
3. Click Prioritize.
4. The App reads the issue's data, sends it to the AI, and displays:
   • The recommended priority level.
   • A 2-3 sentence explanation of why that priority was chosen, referencing your criteria and any matching rules.
   • The current priority for comparison.
5. If you agree with the recommendation, click Apply to update the issue's priority field.
6. If you disagree, simply close the panel. No changes are made.

Using Batch Prioritization

Batch prioritization lets you analyze and reprioritize many tickets at once from the project page.

The flow: Search, Preview, Analyze, Review, Apply

1. Navigate to the AutoPrioritizer AI project page (left navigation in your project).
2. Search for issues using one of the built-in quick filters or by entering custom JQL:
   • Unprioritized -- Issues with no priority set.
   • Backlog -- Issues not assigned to any sprint.
   • To Do -- Issues in "To Do" status.
   • All Open -- All issues that are not Done.
   • Custom JQL -- Write your own query for precise targeting.
3. Preview the matching issues. The App shows a list of issues with their current priority, summary, and type.
4. Analyze -- Click the analyze/prioritize button to send all matching issues to the AI. The App processes issues in batches of 15 for efficiency.
5. Review the results in a table showing: issue key and summary, current priority, recommended priority, AI's reasoning, and whether the priority would change.
6. Apply -- You can apply recommendations individually, apply all at once with "Apply All", or skip any you disagree with.

Setting Up Auto-Prioritize on Issue Creation

Auto-prioritize automatically assigns a priority level to every new issue the moment it is created in a project. No manual intervention is required.

How to enable it

1. Go to the AutoPrioritizer AI project page.
2. Open the Settings/Configure tab.
3. Toggle Auto-prioritize new issues to on.
4. Save your profile.

How it works

• When a new issue is created in the project, a Forge trigger fires automatically.
• The App reads the new issue's data, analyzes it against your project's profile, and sets the priority field.
• This happens in the background, typically within a few seconds of issue creation.
• Auto-prioritized issues count toward your monthly quota.

Managing Custom JQL Quick Filters

Quick filters are saved JQL snippets that appear as one-click buttons on the batch prioritization page.

Default filters

The App comes with four default filters:

• Unprioritized: AND (priority IS EMPTY OR priority = "None")
• Backlog: AND Sprint IS EMPTY
• To Do: AND status = "To Do"
• All Open: AND statusCategory != Done

Each filter is automatically combined with project = "YOUR-PROJECT" and ORDER BY created DESC.

Adding or editing filters

1. Go to the project page and find the filter management section.
2. Add a new filter by providing:
   • Label -- The button text (e.g., "Critical Bugs").
   • JQL suffix -- The JQL clause to append (e.g., AND issuetype = Bug AND priority = Highest).
3. Save your filters.

Understanding the Quota System

How it works

• Every active user gets 100 issues per month.
• An "active user" is anyone who has interacted with the App during the current calendar month.
• Quota is shared across the project, not per-user. If your project has 5 active users this month, the project has 500 issues available.
• Quotas reset on the 1st of each calendar month (UTC).

What counts toward the quota

• Each issue analyzed counts as 1, regardless of whether it is single-issue, batch, or auto-prioritized.
• Searching for issues (without analyzing them) does not count.
• Applying a previously generated recommendation does not count again.

Checking your usage

The project page includes a usage dashboard showing:

• Current month's usage and remaining quota.
• The quota limit and how many active users contribute to it.
• When the quota resets.
• Historical usage trends.

What happens when the quota is exceeded

• Manual prioritization requests (single and batch) will display an error message showing your current usage, limit, and reset date.
• Auto-prioritize silently stops processing new issues until the quota resets.
• Usage resets automatically on the first of the next month.

Troubleshooting

"ANTHROPIC_API_KEY environment variable not set"

The App cannot find your Anthropic API key. This is the most common setup issue.

Fix:

forge login
forge variables set --encrypt ANTHROPIC_API_KEY sk-ant-api03-your-key-here

Make sure you select the correct environment (production for Marketplace installs). After setting the variable, the change takes effect immediately -- no redeployment is required.

"LICENSE_REQUIRED" error

Your Atlassian Marketplace subscription is inactive or has expired.

Fix: Check your subscription status in the Atlassian Marketplace. Renew or reactivate your subscription.

No issues found in batch search

The JQL query returned no results.

Possible causes:

• The project has no issues matching the selected filter.
• All matching issues are in "Done" status (excluded by most default filters).
• Custom JQL has a syntax error -- try the JQL in Jira's issue navigator first to verify it works.

"Quota exceeded" error

You have used all available prioritization credits for the current month.

Fix: Wait for the quota to reset on the 1st of the next month. The error message will show the exact reset date. To increase your effective quota, have more team members use the App -- each active user adds 100 issues to the shared pool.

AI recommendation seems wrong

The AI's recommendations depend heavily on your configuration.

Things to check:

• Are your criteria weights set correctly? A criterion at 0% is completely ignored.
• Is your team context descriptive enough? The more context the AI has, the better its decisions.
• Are your priority level definitions clear? Vague definitions lead to inconsistent results.
• Does the ticket have enough information? A ticket with only a summary and no description gives the AI very little to work with.
• Check if a custom rule is matching unexpectedly and overriding the criteria-based analysis.

Prioritization is slow

The App calls the Anthropic API for each batch of 15 issues. Large batch requests (50-100 issues) may take 15-30 seconds.

Tips:

• Use more specific JQL filters to reduce the number of issues analyzed.
• The single-issue panel is typically faster (under 5 seconds) since it only processes one ticket.

Auto-prioritize is not working

New issues are not getting priority assigned automatically.

Check:

1. Is auto-prioritize enabled in the project's profile settings?
2. Is the Anthropic API key set correctly?
3. Has the monthly quota been exceeded?
4. The trigger only fires on issue creation, not on issue updates or transitions.

Cost Expectations

AutoPrioritizer AI uses Anthropic's Claude Haiku model, which is designed for fast, cost-effective analysis. Typical costs:

• Single issue: Fraction of a cent per analysis.
• Batch of 50 issues: A few cents.
• Typical team (500 issues/month): Well under $1/month in API costs.

You can monitor your Anthropic API spending at console.anthropic.com under the Usage section.

Support

If you run into issues not covered here, contact:

AutoPrioritizer
Email: support@autoprioritizer.app