More review agent refinements

[jasnell]

• Cover rust, js, and ts checklists
• Look for copyright headers
• Add a skill for identifying the current user for better handling in code review and find-owners
• Flag dependency updates in code reviews

[ask-bonk]

The discoverRustCrate function uses substring matching (.includes()), which could match unintended crates. For example, searching for "rand" would match "rand_core", "rand_chacha", etc. This is a known trade-off for fuzzy matching and the function caps results at 3, so it's not a correctness bug.

After reviewing all the changes thoroughly, this PR is well-structured documentation and tooling for AI agent workflows. The changes are:

• New skills for Rust review, TS style, and reviewer identification
• New bazel-deps tool for dependency impact analysis
• Updates to existing commands to support Rust and multi-language workflows
• New commands (dep-impact, rdeps)

There are no runtime code changes, no security implications, no backward compatibility concerns. The code quality of the TypeScript tool is good. The minor issues I found (inconsistent trailing slash on src/pyodide, extra single quotes in bazelQuery) are not actionable in the sense of being bugs.

LGTM

github run