Feature: Implement /insight command for session analysis and personalization

Problem Statement

Currently, OpenHands users lack a built-in mechanism to review historical usage, identify behavioral patterns, or receive automated suggestions for personalizing their environment.

Description

Inspired by the \insight feature in Claude Code, this proposal introduces a keyword-triggered skill to generate data-driven reports from user sessions. The goal is to provide a "meta-analysis" of agent usage and offer actionable insights for improvement.

Functional Requirements

• Trigger: A new keyword-triggered skill activated by the /insight command.
• Data Parsing: Ability to parse saved session data and logs (typically located in the profile or project directory).
• Analysis: The agent analyzes common failures, bottlenecks, or repetitive tasks.
• Report Generation: Generates an HTML report stored in a dedicated local directory (e.g., .openhands/usage-data).
• Optimization Suggestions: Recommends new skills, rules, or configuration changes to streamline the workflow.
• Interactive Implementation: Upon user request, the agent should offer to automatically implement one of the suggested optimizations.

Ideal Workflow (CLI/Web)

• User Input: The user enters /insight.
• Log Scanning: OpenHands scans historical session JSON files.
• Summarization: The LLM identifies usage trends and bottlenecks.
• Report Output: An HTML dashboard is generated and saved locally for the user to review.
• Proactive Suggestion: OpenHands prompts the user: "I noticed you often refactor Python types manually. Would you like me to create a custom skill to automate this?"

Conclusion

This feature introduces a layer of self-optimization to OpenHands. By personalizing the experience based on a user’s specific coding style and project needs, the agent becomes progressively more effective over time.

No implementation file or test is named; start by locating the /insight command or keyword-triggered skill and the saved session JSON/log entry points. Clarify the scope for parsing, analysis, HTML output in .openhands/usage-data, and optional optimization implementation before considering the feature done.