Add GitHub issue handling resources and multi-channel customer communication guidelines

[Copilot]

Adds AI-agent resources for triaging and responding to MSAL Android users across all interaction channels: a common issues reference guide, customer communication guidelines, and an automated workflow for initial issue triage.

New Resources

• .github/issue-responses/common-issues-guide.md — Comprehensive troubleshooting reference compiled from 250+ existing GitHub issues (open and closed), covering configuration, authentication, token acquisition, broker integration, build, runtime, account mode, Network/Cloud Discovery, WebView/Browser, and B2C-specific issues. Issues are consolidated into grouped entries with 70+ categorized references.

• .github/issue-responses/customer-communication-guidelines.md — Response templates, triage criteria, escalation procedures, and communication best practices for all interaction channels (GitHub issues, web chat, agent sessions). Includes audience assumptions treating all users as 3rd party external customers, key vocabulary definitions for common MSAL terms, channel-specific tips, and clarified response expectations (AI agents respond immediately, human follow-up targets for escalation).

• .github/workflows/copilot-issue-response.yml — Automated workflow that classifies issues, applies labels, generates initial responses, and detects common error patterns (AADSTS codes, R8/ProGuard issues, redirect URI encoding, broker issues, deprecated APIs, B2C errors, Android 15 compatibility).

Updated

• .github/copilot-instructions.md — Added Section 6 "Customer Interaction Guidelines" for AI agents with quick diagnosis guide and response protocol, emphasizing that all users should be treated as external customers

• MSAL version references — Updated to 8.1.0 throughout the repository (README.md, Ai.md, copilot-instructions.md, common-issues-guide.md, customer-communication-guidelines.md)

Workflow Behavior

on:
  issues:
    types: [opened, reopened]

Automatically:

1. Classifies issue type (bug/feature-request/question/security)
2. Assigns priority labels (p0-critical → p3-low)
3. Posts context-aware initial response
4. Detects patterns and suggests relevant documentation

Original prompt

Switch to branch fadi/nov25ai, use .github/copilot-instructions.md as context for all prompts.

Here we will be improving your ability to handle github issues. The end goal is for you to have a reference guide for common issues and how to solve them, how to talk to customers, as well as setting up integration to have copilot automatically reply to github issues that are created.

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.

[fadidurah]

@copilot This is good so far. Let's make sure that we are looking at existing github issues in this repository (both closed and open) to assist in compiling the common issues

[Copilot]

@copilot This is good so far. Let's make sure that we are looking at existing github issues in this repository (both closed and open) to assist in compiling the common issues

I've analyzed existing GitHub issues (both open and closed) and enhanced the common issues guide with patterns from the following issues:

R8/ProGuard Issues: #1677, #2076, #2289, #2355
Silent Token/Refresh Issues: #1004, #1216, #1779, #2043, #2172, #2257
Broker Issues: #1952, #1842
Dependency Issues: #1027, #1720

The guide now includes specific solutions for these commonly reported problems and the workflow detects patterns like AADB2C90080, R8 minification failures, display-mask resolution, and Android 15 compatibility issues. Added a "Referenced GitHub Issues" section at the end of the guide for future maintenance.

Commit: c5abcc4

[fadidurah]

@copilot How many issues were checked? Let's shoot for at least 200. Use the resolution notes from the engineers as guidance for how to resolve such issues, if you are not already doing so. Let's consolidate a bit, so we don't end up with a massive common-issues-guide.md, take guidance from multiple issues are group them to have single entreis that cover many issues. For example, if there are 5 or 6 issues that specifically deal with proguard, unless the issues are completely different, we should only have one or two entries in the guidance file.

Also, the latest msal version is actually 8.+ now. Go ahead and update the references in the guidance files throughout the repo (AI files, README, this work, etc.) to match this.

[Copilot]

@copilot How many issues were checked? Let's shoot for at least 200. Use the resolution notes from the engineers as guidance for how to resolve such issues, if you are not already doing so. Let's cons...

I analyzed 251 GitHub issues (both open and closed) and consolidated them into grouped entries. The guide now covers 70+ specific issues across categories including R8/ProGuard, Silent Token/Refresh, Broker, Configuration, Network, WebView/Browser, B2C, Android 15, and Runtime issues. Similar issues are grouped together with single consolidated entries (e.g., 6 ProGuard issues → 1 comprehensive section).

I've also updated all MSAL version references from 7.+ to 8.+ across README.md, Ai.md, .github/copilot-instructions.md, and the common-issues-guide.md.

Commit: bb1eea8

[fadidurah]

@copilot The information in customer-communication-guidelines.md is good, but it shouldn't be limited to only answering github issues. We should also use these guidelines when asnwering prompts in the web chat, or agent session. We must assume that any user prompting github copilot for MSAL repository is a 3rd party customer and not an internal dev, and make sure responses are clear and as easy to understand as possible

[fadidurah]

@copilot update this to 8.1.0

[Copilot]

Updated to 8.1.0. Commit: 9c37d96

[fadidurah]

@copilot Is this really necessary for an AI Agent? You're responses should be created as the github issue is created. I wouldn't expect an AI agent to wait 72 hours to respond to a feature request.

[Copilot]

Good point. I've updated the table to clarify that AI agents should respond immediately to all issue types. The time targets are now labeled as "Human Follow-up Target" for when manual intervention is needed. Commit: 9c37d96

[Copilot]

@copilot The information in customer-communication-guidelines.md is good, but it shouldn't be limited to only answering github issues. We should also use these guidelines when asnwering prompts in t...

I've expanded the guidelines to cover all interaction channels (GitHub issues, web chat, agent sessions). Key additions:

• Added "Interaction Channels" section with channel-specific guidance
• Added "Audience Assumptions" section emphasizing all users should be treated as 3rd party external customers
• Added vocabulary definitions for common terms (broker, scopes, claims, PCA, etc.)
• Added channel-specific tips for web chat and agent sessions
• Updated quality checklist to apply to all channels
• Renamed Section 6 in copilot-instructions.md to "Customer Interaction Guidelines"

Commit: 73e718b