kimi-ui-advisor — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited kimi-ui-advisor (Agent Skill) and scored it 96/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 1 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 1 flagged
The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Use Kimi Code CLI as a bounded frontend/UI consultant, not as the file editor. Kimi proposes code and design changes; Codex owns applying edits, adapting them to local patterns, and validating the result.
$kimi-ui-advisor.validation commands.
advise: targeted implementation advice and code suggestions.audit: ranked UI issues before editing.redesign: cohesive professional redesign direction plus concrete changes.component: component API, variants, states, styling, and accessibility.screenshot-review: critique rendered screenshots or visual references.compare: before/after visual QA after Codex applies changes. python3 skills/kimi-ui-advisor/scripts/kimi_ui_advisor.py \
--work-dir "$PWD" \
--mode component \
--file src/components/Button.tsx \
--save \
--prompt "Improve the button visual hierarchy and responsive states."For screenshot review:
python3 skills/kimi-ui-advisor/scripts/kimi_ui_advisor.py \
--work-dir "$PWD" \
--mode screenshot-review \
--image /tmp/dashboard-mobile.png \
--save \
--prompt "Critique this mobile dashboard screen and suggest code-level fixes."For before/after QA:
python3 skills/kimi-ui-advisor/scripts/kimi_ui_advisor.py \
--work-dir "$PWD" \
--compare \
--before-image /tmp/before.png \
--after-image /tmp/after.png \
--save \
--prompt "Identify regressions and remaining polish gaps."style, data contracts, accessibility, and design-system ownership intact.
run browser or screenshot checks when available.
which omits shell and write tools.
accessibility references, and UI library documentation. Do not ask Kimi to search using proprietary source snippets or private requirements.
fixes, infrastructure, dependency upgrades, or broad architecture unless the UI surface depends directly on that context.
CLI setup before relying on the response.
scripts/kimi_ui_advisor.py: deterministic wrapper around kimi --print.assets/kimi-agent/agent.yaml: read/search-only Kimi custom agent.templates/design-brief.md: optional structured brief for high-stakes UI work.references/advanced-modes.md: mode selection and prompt patterns.references/kimi-cli-integration.md: current Kimi docs/source notes.references/output-contract.md: JSON contract and application rules.~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.