Exhaustively dogfood Kovo, file or triage bugz/papercuts ledgers, then implement the fixes through verified parallel worktree batches. Use when asked to dogfood and fix, handle remaining dogfood reports, continue after bugz/papercuts reports, implement newly found Kovo issues, parallelize dogfood remediation, push fixes, keep going until no new dogfood issues remain, or monitor CI until the reports no longer expose open issues.
Dogfood Kovo by scaffolding or running real apps from the local framework, exercising ordinary AND advanced user workflows, diagnosing framework and starter-template papercuts, and writing the next plans/papercuts ledger. Use when asked to "dogfood" Kovo, try create-kovo end to end, exercise advanced features, reproduce rough edges from app usage, or document papercuts without immediately fixing them. Scales from a single quick app to an exhaustive multi-app workflow-orchestrated sweep with adversarial verification.
Run a fresh adversarial bug hunt across the Kovo framework and produce the next dated severity-ranked bug ledger at plans/bugz-N.md. Use when asked to "find bugs", "audit the codebase for bugs", "do a security/correctness sweep", "create a bugz report", or otherwise hunt for NEW, distinct, reproduced defects (not to fix them). Each finding is reproduced in a throwaway git worktree via a multi-agent find → adversarial-verify → completeness-critic loop. This skill finds and reports; it does not fix.
Complete a specified Kovo active plan end to end in a new git worktree, track the effort with a Codex goal, coordinate sub-agent workers for large phase-sized disjoint implementation chunks, reject tiny proof-slice delegation, integrate and verify worker commits in the main agent worktree, push substantial batches, and monitor CI. Use when asked to "implement this plan", "finish plans/foo.md", "complete the roadmap item", or otherwise execute an active `plans/*.md` implementation ledger rather than merely review or update it.
Restore Kovo's `main` branch to green after the latest pushed commit has failing GitHub Actions checks. Use when asked to inspect the latest GitHub commit, debug failing CI or GitHub Pages runs, repair workflows or code, optionally do the repair in a separate worktree, commit and push the fix, and monitor the next Actions results.
Use when the user wants to design, redesign, shape, critique, audit, polish, clarify, distill, harden, optimize, adapt, animate, colorize, extract, or otherwise improve a frontend interface. Covers websites, landing pages, dashboards, product UI, app shells, components, forms, settings, onboarding, and empty states. Handles UX review, visual hierarchy, information architecture, cognitive load, accessibility, performance, responsive behavior, theming, anti-patterns, typography, fonts, spacing, layout, alignment, color, motion, micro-interactions, UX copy, error states, edge cases, i18n, and reusable design systems or tokens. Also use for bland designs that need to become bolder or more delightful, loud designs that should become quieter, live browser iteration on UI elements, or ambitious visual effects that should feel technically extraordinary. Not for backend-only or non-UI tasks.
Audit Kovo's manifest-declared public API surface from a conservative JavaScript framework-author perspective. Use when asked to scrutinize public exports, decide whether APIs should remain public, narrow/remove exposed symbols or subpaths, compare API shape against SPEC.md and repo examples, or generate a dated public API audit report under plans/.