Background
During review of PR #56 (add /ai-coding-config doctor), claude[bot] noted a design smell:
The sync comment at line 561 acknowledges: 'These checks mirror the ones in <marketplace-doctor> (update-mode). If either section is updated, keep both in sync.'
This is the right thing to note, but it's a maintenance trap. Consider having <doctor-mode> explicitly reference <marketplace-doctor> with an instruction like "run the same plugin install state checks as in <marketplace-doctor>" — the LLM can read the sibling section rather than requiring manual sync.
Also flagged: no agent frontmatter validation in doctor. If agents have required fields (name, description, color), broken agent files are invisible to doctor. Worth adding or explicitly calling out as out of scope.
Proposed work
- Refactor
<doctor-mode> to reference <marketplace-doctor> by name rather than duplicating the check list
- Evaluate adding agent frontmatter validation to doctor (or document why it's out of scope)
Surfaced by PR #56 review sweep (PR #59).
Background
During review of PR #56 (add /ai-coding-config doctor),
claude[bot]noted a design smell:Also flagged: no agent frontmatter validation in doctor. If agents have required fields (name, description, color), broken agent files are invisible to doctor. Worth adding or explicitly calling out as out of scope.
Proposed work
<doctor-mode>to reference<marketplace-doctor>by name rather than duplicating the check listSurfaced by PR #56 review sweep (PR #59).