1 Goal
2 Coding (JS Script)
3 Configuring (JS Listener)
4 Testing
5 Improving
- 5.1 Configuration Management
- 5.2 Jira Comment Attribution

Goal

Next, we will develop a practical example: an issue priority classifier for newly created issues: JavaScript Runner will listen for issue.create events and trigger a LangChain script to prioritize the issue based on its description, using natural language processing to analyze context and urgency.

Coding (JS Script)

Copy this script below in the JS Console
Provide your OpenAI API Key,
Save it with the “issue_priority_classifier” name:


import { OpenAI } from "@langchain/openai";

const issueId = globalThis.__issueid; //The newly created issue id


const JIRA_BASE_URL = jira.getBaseUrl();

const OPENAI_API_KEY = ""; // <- Provide yours

const httpHeader = {
    Accept: "application/json",
    "Content-Type": "application/json",
};

async function getIssue(issueId) {
    const res = await fetch(
        `${JIRA_BASE_URL}/rest/api/2/issue/${issueId}?fields=summary,description,priority`,
        { headers: httpHeader }
    );
    if (!res.ok) throw new Error(`Failed to fetch issue: ${res.status} ${res.statusText}`);
    const { fields } = await res.json();
    return fields;
}

async function getPriorities() {
    const res = await fetch(
        `${JIRA_BASE_URL}/rest/api/2/priority`,
        { headers: httpHeader }
    );
    if (!res.ok) throw new Error(`Failed to fetch priorities: ${res.statusText}`);
    return await res.json();
}

async function updatePriority(issueId, priorityId) {
    const res = await fetch(
        `${JIRA_BASE_URL}/rest/api/2/issue/${issueId}`,
        {
            method: "PUT",
            headers: httpHeader,
            body: JSON.stringify({ fields: { priority: { id: priorityId } } }),
        }
    );
    if (!res.ok) throw new Error(`Failed to update priority: ${res.statusText}`);
}

async function addComment(issueId, commentBody) {
    const res = await fetch(
        `${JIRA_BASE_URL}/rest/api/2/issue/${issueId}/comment`,
        {
            method: "POST",
            headers: httpHeader,
            body: JSON.stringify({ body: commentBody }),
        }
    );
    if (!res.ok) throw new Error(`Failed to add comment: ${res.statusText}`);
}


async function classifyPriority(summary, description) {
    const llm = new OpenAI({ openAIApiKey: OPENAI_API_KEY });
    const prompt = `You are an assistant that assigns a Jira ticket priority.
Available levels: Blocker, Highest, High, Medium, Low, Lowest, Minor.
Summary: ${summary}
Description: ${description}

Respond with JSON:
{
  "new_priority": "<level>",
  "reasons": "<brief explanation>"
}`;

    const response = await llm.invoke(prompt);

    const text = response.trim();

    try {
        return JSON.parse(text);
    } catch (err) {
        throw new Error(`Unable to parse OpenAI response: ${text}`);
    }
}

try {
    const { summary, description, priority } = await getIssue(issueId);


    const { new_priority, reasons } = await classifyPriority(summary, description);


    if (new_priority !== priority.name) {

        const priorities = await getPriorities();
        const target = priorities.find(p => p.name.toLowerCase() === new_priority.toLowerCase());
        if (!target) throw new Error(`Priority '${new_priority}' not found in Jira`);

        await updatePriority(issueId, target.id);


        const comment = `Priority changed from **${priority.name}** to **${new_priority}**.

**Reasons:** ${reasons}`;
        await addComment(issueId, comment);
    } else {
        //TThe current priority already matches the suggestion. No changes made.
    }

} catch (err) {
    jira.log.warn("Failed issue priorization: " + err);
}

Configuring (JS Listener)

Navigate to:

Administration → Manage apps → JavaScript Runner → JS Listener

Click Create Listener and fill out the pop up dialog:

Name: issue_created_classifier
Event: issue.created

Click Accept to save the new JS Listener.

Testing

Test it! Create a new issue with this description:

I've seen this error in the system log. It does not seems to much relevant as the system is working perfectly.


2025-05-29T12:47:05.842Z [ERROR] [token-service:production] trace_id=9f1c2a7d3b4e5f6081a2bc3d4e5f6a7b span_id=7a6b5c4d3e2f1a0b
    host=auth01.prod.internal user_id=98765 client_ip=203.0.113.42
    → JWT signature verification fallback triggered (expected=RS256, got=HS256, kid=main-rsa-key)
       Insecure JWT validation: signature verified with HMAC instead of RSA
       Note: public key used as HMAC secret – potential token forgery

And set Low priority:

Open the new issue and check the priority and comments section:

The LangChain script has been triggered and it changed the priority from Low to High.
Furthermore a new comment explaining the reason of the change has been added to the issue!

Improving

Configuration Management

To improve maintainability:

Move the OpenAI API key to a dedicated configuration module in Jira.
Import the API key into the LangChain script, ensuring that updates to the key require changes in only one location.
Encapsulate the OpenAI model name in the same module to handle potential model deprecation by OpenAI, minimizing script modifications.

Jira Comment Attribution

To avoid confusion in production:

The current setup attributes comments to the issue reporter (current user), which may mislead them into thinking they created the comment.
Create a dedicated Jira “bot” user for automated tasks.
Use the bot’s credentials in the create-comment function to ensure that comments and priority changes are attributed to the bot, enhancing transparency for the reporter.

AI Classifier for issue priorities