niekcandaele

Manual E2E tester that starts the app and exercises new features end-to-end

Generate a single self-contained HTML page that is genuinely visual AND interactive — charts, diagrams, motion, tabs, comparison toggles, click-to-expand, base64-inlined images, opinionated typography. Use whenever the user wants ANY rich visual artifact from arbitrary content: explainer, research write-up, PRD or spec page, pitch, internal one-pager, "make this less boring" rebuild, scroll-snap deck, landing-style summary, distilled report. Trigger phrasings: "make me a page about X", "turn this PDF/doc into something visual", "build me a deck/talk/pitch", "explain Y in a visual way", "make this readable", "give it some eye candy", "I want something I can show the team", "less boring version of this", "rebuild that page", "redo it with more visuals". Output is one .html file that renders identically when DM'd — CDN libraries (Tailwind, Chart.js, D3, GSAP, Mermaid, Lucide) load from stable jsdelivr/unpkg URLs; every image is base64-inlined. Not for plain Markdown docs (use technical-writer), not for code revi

Vision-based visual QA reviewer — captures rendered output (live web pages, static HTML artifacts, PDFs) as screenshots, inspects them with a designer's eye for layout defects a human catches instantly, and normalizes findings into the verify pipeline format

Independent second-opinion reviewer that shells out to the local Codex CLI for a broad code review, then normalizes findings into the verify pipeline format

Comment-hygiene-only reviewer — flags ephemeral review-ID references, historical change-narration, stale comments, reviewer-appeasement, and redundant restating in the scoped diff, and normalizes findings into the verify pipeline format

Adversarial cooperation loop — player implements, /verify reviews, creates PR, passes CI

Comprehensive code reviewer combining design review, architecture, coherence, hardening, and security analysis

Run comprehensive triage-first verification pipeline with specialized skills. Detects scope, discovers toolchain, triages files to relevant skills, runs static analysis, invokes review skills in parallel, exercises the app, and produces a unified report. Supports interactive, report-only, and auto-fix modes.

Generate a rich HTML insights report from adversary CLI run artifacts on disk. Use this whenever the user asks to analyze their adversary runs, see how adversary has been performing, generate a run report, look at convergence patterns across runs, find friction in adversary usage, evaluate whether the verify pipeline is healthy, or figure out which repos/plans work well with adversary — even if they don't explicitly say "insights" or "report". Triggers on phrases like "analyze my adversary runs", "show me how adversary has been doing", "make me a report from my recent runs", "is adversary working well in this repo", "what's going wrong in my adversary runs", and similar. Reads ~/.local/state/adversary/, extracts per-run quality facets, aggregates cross-repo, and renders a self-contained HTML dashboard with a per-repo report card, charts, narrative analysis, and click-to-expand evidence linking back to source artifacts.