Hostile, bias-free review of a change — dispatches independent adversarial reviewers that hold the code against what was agreed and question whether what was agreed was right, plus a standalone false-positive filter, and reports only verified findings in plain language. Ships a full 18-reviewer panel, a cheaper quick panel that proposes the fewest reviewers that fit the change, lets you add or drop any, and keeps both verification gates, and a fast one-reviewer review that checks its own findings and re-checks only what changed since the last review.
3 skills
Keeps what the agent makes from reading as AI slop, in two skills. anti-slop-ui covers interfaces: no gradient-everything, rainbow palettes, pulsing badges for states that never change, fingernail-striped cards, emoji icons, default fonts, uppercase labels, scroll effects, screens with no hierarchy or no empty, loading, and error states, or landing-page hype on working screens. anti-slop-text covers prose: no invented details, inflated significance, promotional tone, weasel attributions, not-X-but-Y contrasts, manufactured rhythm, AI vocabulary, chat residue, restating conclusions, or PR text about what was left unchanged. Each checks its output against a catalog of known tells before it ships.
2 skills
Fully local visual plans and recaps — the agent writes markdown, a bundled zero-dependency server renders it in your browser with Mermaid diagrams, rich diffs, styled DB migrations, API call cards, and a read-only OpenAPI view, with live reload and a comment feedback loop. No remote services.
2 skills
Comment discipline for code the agent writes — every comment stays within a line or two, carries information the code does not, and describes the current state only, with no iteration labels, finding numbers, wave or task IDs, project phase names, or narration of how the code got here; regressions are pinned by tests instead.
1 skill
Re-asks the questions an agent has open, one per chat message, in one fixed shape — a numbered claim, the situation, a numbered flow to the consequence, the fix, its cost, your call — after checking the facts and that each recommended fix really works, and builds nothing until every question has a direct apply, reject, or different direction. After a doubtful reply it asks the user to say which they mean. Never uses the harness question tool.
1 skill
Language-agnostic code simplification — reduces complexity (nesting, long functions, duplication, dead code, unclear names) while preserving exact behavior, applying one tested change at a time within a deliberate scope.
1 skill