菜单
Respond to Bouts crises including methodology criticism, challenge exploits, AI lab complaints, downtime, and negative press with specific response templates, escalation rules, and the principle of always responding with data and transparency instead of defensiveness.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Internal admin tools for inspecting, monitoring, and improving judge outputs — including raw output inspection, calibration drift detection, missing evidence detection, low-signal output flagging, and a feedback quality dashboard.
Block filler, detect low-signal text, score specificity, and enforce evidence-anchored observations across all LLM-generated feedback in Bouts — with banned phrase detection, specificity scoring, retry-with-critique, and dimension name normalization.
Generate insights comparing an agent's performance to top performers, peers, and their own history — only from real data with minimum sample guards, including percentile comparisons, counterfactual rank calculation, and "surprisingly strong" lane detection.
Evaluation-specific visualization patterns for Bouts — radar charts, multi-judge comparison bars, rank distribution dots, confidence overlays, and percentile bands using Recharts with full accessibility and missing-data handling.
What users actually need after losing or winning a competitive AI evaluation — emotional state design for winners, close misses, and clear losses, with TSX layout and copy patterns that make feedback feel fair rather than algorithmic.
Expose when a judgment is high-confidence vs thin-evidence without undermining the platform's authority — covering the confidence trilemma, per-tier UI patterns, data model, copy patterns, and hard rules on when NOT to show confidence indicators.
| name | crisis-communications |
| description | Respond to Bouts crises including methodology criticism, challenge exploits, AI lab complaints, downtime, and negative press with specific response templates, escalation rules, and the principle of always responding with data and transparency instead of defensiveness. |
Never attack. Acknowledge fast. Be specific. Show data. Fix publicly.
"We take this seriously. Here's the data: [specific numbers]. Our judges use 3+ different model families, score independently, and are calibrated weekly against known-quality benchmarks. If you have specific evidence of bias, we want to see it — submit via our Fair Play system."
"We identified an exploit in [challenge]. It's been quarantined. Affected scores are being reviewed. The fix: [specific change]. This is why we track exploit resistance as a CDI component — when challenges show vulnerability, we catch it."
Respond privately first: "We'd love to review the specific runs you're concerned about. Our disputes process includes full audit trails. Let's look at the data together." Never engage in public disputes with potential customers.
"Bouts is currently experiencing [issue]. ETA for resolution: [time]. Challenge submissions during this window will not be affected — we'll rerun any impacted submissions. Status updates every 30 minutes at [status page]."
Contact the journalist directly with specific factual corrections and data. Never attack publicly. Correct privately first. If they won't correct, publish your own response with data.
Show the judge diversity data and calibration methodology. Offer a public audit. "Transparent methodology is our whole value proposition — we welcome scrutiny."
