Conflicts: install one, not both

Loading both of these gives the agent two processes for one job. The argument then happens inside its context, where you cannot see it, and the result is whichever instruction happened to weigh more.

Pair	Why
test-driven-development tdd	Two complete test driven development processes for the same job. Loaded together they hand the agent competing instructions about cycle size and when to refactor. Pick one.
systematic-debugging diagnosing-bugs	Both replace guess and patch with a staged diagnosis, and they stage it differently. Running both produces a process argument inside the agent's context, where you cannot see it.
superpowers tdd	The superpowers methodology already contains its own test driven development skill. Adding a second one is not redundancy, it is a contradiction the agent has to resolve on its own.
superpowers diagnosing-bugs	Same collision on the debugging side. superpowers ships systematic-debugging, so a second debugging methodology competes with it rather than adding to it.

Pairings: each covers the other's gap

These are not simply skills that work together. In each pair, one skill has a hole that the other fills, which is why installing only one produces a predictable disappointment.

Pair	Why
verification-before-completion webapp-testing	The strongest pairing in the catalogue. One makes evidence mandatory before an agent claims a task is done, the other is what actually produces that evidence in a browser.
writing-plans subagent-driven-development	Splitting work across agents only survives contact with reality when a written plan defines the boundaries first. Reverse the order and you get parallel versions of the same misunderstanding.
brainstorming grill-me	Brainstorming widens the option space, grill-me attacks the option you chose. Used in that order they catch different failures: a missing alternative and an undecided branch.
frontend-design theme-factory	frontend-design decides how one page should look, theme-factory makes that decision reusable. Without the second, page ten looks nothing like page one.
brand-guidelines theme-factory	Guidelines state the rules a brand lives by, a theme is the implementation those rules produce. Teams that install only the first keep re-explaining the same palette.
doc-coauthoring docx	One structures the writing, the other produces the file. On their own you get either an unreviewable draft or a well formatted document nobody agreed on.
brand-guidelines pptx	Decks are where visual drift shows first. The deck skill builds the structure, the brand skill is what stops every deck from inventing its own spacing.
skill-creator find-skills	Search before you write. Installing the discovery skill first is what stops skill-creator being used to rebuild something that already exists.
mcp-builder claude-api	Building a server and calling the model are the two halves of the same integration. The API reference is what keeps the identifiers and call patterns current while the server is designed.
web-artifacts-builder frontend-design	The artifact skill decides that the page runs on its own, the design skill decides that it does not look like every other generated page. Neither covers the other.

Methodology

A conflict is recorded only when both skills instruct the agent on the same decision, not merely when they cover related subjects. A pairing is recorded when each one covers a gap the other leaves open. Every claim below comes from reading both skills, and the review date is the one shown on each entry.