refactor: dev to main - Update azure.yaml with GitHub issue URL

[Pavan-Microsoft]

Purpose

This pull request introduces improvements to the deployment experience and refines the logic for generating charts in agent scripts. The main changes are the addition of user guidance during provisioning and stricter rules for chart generation to prevent unintended behavior.

Deployment experience improvements:

• azure.yaml: Added preprovision hooks for both Windows and POSIX environments to display a helpful message if deployment fails, directing users to open an issue with deployment logs.

Agent behavior updates:

• infra/scripts/agent_scripts/01_create_agents.py: Updated chart generation instructions to prevent the agent from inventing new queries or defaulting to generic topics when the user request lacks a specific metric or subject. The agent must now use the most recent prior numeric dataset in the conversation, or return an explicit error if none exists.

Does this introduce a breaking change?

• Yes
• No

Golden Path Validation

• I have tested the primary workflows (the "golden path") to ensure they function correctly without errors.

Deployment Validation

• I have validated the deployment process successfully and all services are running as expected with this change.

[Copilot]

Pull request overview

Adds a Windows-specific preprovision hook to azure.yaml so azd shows a prominent, colored message pointing users to the project’s GitHub issues page before provisioning begins, helping users know where to report deployment problems.

Changes:

• Added a hooks.preprovision.windows PowerShell script that prints a highlighted “open an issue” notice and issues URL.
• Enabled interactive execution for the new Windows hook.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

Comments suppressed due to low confidence (1)

infra/scripts/agent_scripts/01_create_agents.py:60

• The new chart rule says to take the numeric dataset from the “most recent prior answer”, but later the prompt requires that numeric data must not be shown in the final answer. In cases where the prior response was a non-chart answer, the numeric dataset will not be present in that answer, making this instruction internally inconsistent and likely to cause the model to hallucinate data/query context.

        - If the chart request does not itself specify a metric or subject to visualize (e.g., a bare "generate chart", "create a chart", "make a graph"), you must NOT invent a new or unrelated query or default to any topic (such as top topics by conversation). Instead, take the subject and numeric dataset from the **most recent prior answer in THIS conversation** and chart exactly that data, re-running the same SQL query if needed to obtain it. Only if this conversation has no prior answer containing a numeric dataset, return exactly: {"error": "Chart cannot be generated"}.
        - If no numeric dataset is available from either the current query or previous context, return exactly: {"error": "Chart cannot be generated"}.
        - Do not invent or rename metrics, measures, or terminology. **Always** use exactly what is present in the source data or schema.
        - When the user requests a chart, the final response MUST be the chart JSON ONLY.
        - Numeric data must be computed internally using SQL, but MUST NOT be shown in the final answer.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

Comments suppressed due to low confidence (1)

infra/scripts/agent_scripts/01_create_agents.py:56

• This PR’s title/description focus on adding an azure.yaml preprovision hook, but this change also updates the agent instruction prompt for chart behavior. Please either update the PR description/title to include this behavior change or move this prompt change into a separate PR to keep scope clear.

        - If the chart request does not itself specify a metric or subject to visualize (e.g., a bare "generate chart", "create a chart", "make a graph"), you must NOT invent a new or unrelated query or default to any topic (such as top topics by conversation). Instead, take the subject and numeric dataset from the **most recent prior answer in THIS conversation** and chart exactly that data, re-running the same SQL query if needed to obtain it. Only if this conversation has no prior answer containing a numeric dataset, return exactly: {"error": "Chart cannot be generated"}.