assumptions

Turn the current conversation into a spec — no interview, just synthesis of what you've already discussed.

Diagnosis loop for hard bugs, regressions, flaky failures, and performance problems.

Authoritative, source-cited reference for prompt engineering and context engineering with current frontier LLMs (Claude Opus 4.7 / Sonnet 4.6, GPT-5.4 / 5.5…

Senior, sceptical, hands-on exploratory QA of a running application: web apps (including data-heavy business tools), desktop apps, and terminal TUI or CLI…

Turn this shell test suite into a replayable corpus, or seed a corpus from tests/<x>.test.sh with pending rows for the known gaps.

Audit a whole repo or module for over-engineering that is alive and in use — speculative generality, pass-through layers, frameworks where a function would…

Sweep a repo or directory for accumulated debt — TODO/FIXME/HACK markers, dead code, obsolete workarounds, stale dependencies — and hand back a delete-ledger…

Implement a piece of work from a plan, a spec or a set of tickets, one task at a time with review between steps.

Shrink a feature request to the smallest version that still achieves its goal before any code is written.

Report the terse mode level that the verbosity governor enforces (off, on, hard) and where it is set.

Write a visual, single-file HTML onboarding report for the repo in the working directory, in the design system the plugin's design_system setting names…

Use on /1337:tdd, or when the user wants to build features or fix bugs test-first, mentions "red-green-refactor", or wants integration tests.