Explore workspace-aware Agent Mode for more accurate error explanations

[donhui]

Problem

What problem does this solve?

The plugin currently explains failures mainly from Jenkins logs and optional custom context. While this is useful, it can be insufficient when the root cause depends on repository or workspace context that is not visible in logs alone.

Examples include build, test, or pipeline failures where understanding related source files, project configuration, or pipeline definitions would lead to a more accurate explanation.

Solution

Describe your proposed solution:

Explore an optional Agent Mode that allows the plugin to use limited workspace context as part of failure analysis, in addition to build logs.

At a high level, this could help the plugin produce more repository-aware and actionable explanations by considering relevant files from the workspace when needed.

This issue is intentionally scoped as a discussion starter. The detailed design, boundaries, and implementation approach should be defined collaboratively in follow-up discussion.

Alternatives

What alternative approaches have you considered?

• Continue with log-only analysis and improve prompting
• Rely more on manually provided custom context
• Investigate other repository-aware approaches that are less agentic

These may help in some cases, but may still fall short when workspace context is important to understanding the failure.

Additional Context

Screenshots, examples, or other relevant information:

This issue is meant to gauge interest and align on direction before proposing a detailed implementation.