investigate code review modes

in looking at https://github.com/intercom/2x-skills/blob/main/plugins/code-review-tools/skills/thermo-nuclear-code-review/SKILL.md, from the article at https://hbr.org/2026/03/when-using-ai-leads-to-brain-fry, i see a need to make the code review skill re-structured in a way that allows different "modes" of review. the intent is to have different focus areas for the review, rather than just a size variance in determining which agents and how many agents to send into the review

ideas for modes include:

• general (current "mode")
• architecture
• data,
• object-oriented languages
• functional languages
• language-specific modes
• mode to look at specific tickets and ensure the work fits the ticket requirements
• etc

the idea for implementation is centered around reference files, with a new file for each mode. the high level skill itself becomes a workflow that determines the mode, loads the mode specifics, executes the review in that mode, and synthesizes the final report

han config.md must support custom modes so projects, teams, etc, can provide and share custom modes between engineers

Start by reading plugins/code-review-tools/skills/thermo-nuclear-code-review/SKILL.md and han config.md, then trace how the current review mode selects agents and synthesizes its report. Done means a concrete design for mode-specific reference files, mode selection, review execution, final synthesis, and custom modes in configuration.