What problem does it solve? Before shipping a Convex backend, teams run separate authz, code-review, performance, and log audits and end up with four disconnected reports, duplicated findings, and no clear priority order. This Skill composes all audit passes into a single scored readiness report with a deduplicated, ranked fix plan. ## Core Features & Use Cases - Composed audit passes: Runs convex-authz, convex-reviewer, convex-advisor, and convex-insights, skipping passes whose preconditions (deployment, traffic) are unmet and stating what was skipped. - Auditable scoring: Starts at 100 and subtracts per confirmed finding by severity (high −15, med −5, low −1), printing the exact formula and per-class breakdown; plausible-only findings never move the score. - Deduped findings with fix dispatch: Normalizes each finding to a function/table identity so the same defect seen from code and deployment loci collapses to one, then dispatches accepted fixes to the owning fixer capability and re-runs passes to show the score delta. - Use Case: Before promoting a Convex app to production, run the readiness report to get a score, a blockers/should-fix/nice-to-have breakdown, and an ordered plan of which fixer capabilities to invoke next. ## Quick Start Run a launch-readiness report on my Convex backend and give me the score with an ordered fix plan.
