「judge」の検索結果

Triage a single GitHub PR by PR number. Check out the PR's branch, detect conflicts with the target branch via `gh pr status` (and label the PR with `cc-resolve-conflict` if any are found), collect unresolved review comments and CI status and judge only whether each item must be fixed, then take action (add cc-fix-onetime label if fixes are needed; if release-ready, add cc-release-ready label for an Epic PR marked `cc-epic-issue` instead of merging, otherwise merge the PR).

日本語の概要は準備中です。原文の説明を表示しています。

Detects and removes contamination in genome assemblies via two disjoint workflows - foreign-sequence screening of a single-organism (eukaryote/isolate) assembly with NCBI FCS-GX (GenBank-submission-mandatory), FCS-adaptor, and BlobToolKit blob plots; and MAG/bin quality assessment with CheckM2 plus GUNC (chimerism) plus GTDB-Tk taxonomy, judged against MIMAG. Covers why CheckM2 alone is blind to disjoint-marker chimeras, the FCS-GX RAM wall, organelle/NUMT triage, strain heterogeneity, and the HGT-vs-contamination (tardigrade) trap. Use when screening an assembly for foreign contamination before GenBank submission, assessing MAG completeness/contamination/chimerism, deciding which contigs to remove, or distinguishing real HGT from contaminant contigs.

日本語の概要は準備中です。原文の説明を表示しています。

Spec-locked TDD with a mutation-testing gate — refine the intent into an immutable Gherkin contract (SHA-256 frozen), implement under the three laws of TDD, pass a judge audit, and merge only if every mutant dies.

日本語の概要は準備中です。原文の説明を表示しています。

This skill should be used when the user asks to "implement LLM-as-judge", "compare model outputs", "create evaluation rubrics", "mitigate evaluation bias", or mentions direct scoring, pairwise comparison, position bias, evaluation pipelines, or automated quality assessment.

日本語の概要は準備中です。原文の説明を表示しています。

Review, repair and steward the nightly Codex debt-sweep PR (branch codex/daily-debt/<date>). Judges every commit GOOD / REPAIR / REVERT, triages the bot review findings through /fix, lands one round-1 commit, files follow-up issues, then stewards the PR itself with a 3-round ceiling. Run by the 08:00 UTC cloud routine; also '/debt-review', '/debt-review 1789'.

日本語の概要は準備中です。原文の説明を表示しています。

Cross-model benchmark for gstack skills. Runs the same prompt through Claude, GPT (via Codex CLI), and Gemini side-by-side — compares latency, tokens, cost, and optionally quality via LLM judge. Answers "which model is actually best for this skill?" with data instead of vibes. Separate from /benchmark, which measures web page performance. Use when: "benchmark models", "compare models", "which model is best for X", "cross-model comparison", "model shootout". (gstack) Voice triggers (speech-to-text aliases): "compare models", "model shootout", "which model is best".

日本語の概要は準備中です。原文の説明を表示しています。

UI 預覽評審 SKILL — 截取 Cocos Editor Preview 畫面，與 proof draft 提供的參考圖逐 zone 比對，輸出視覺信心分數與 PASS/FAIL 評審報告，並回寫 runtime-verdict。USE FOR: ui-asset-qc 通過後，需要人工確認 + 自動評審的最終視覺驗收。DO NOT USE FOR: Runtime 崩潰（用 cocos-log-reader）、Browser QA 截圖（用 cocos-preview-qa）、資產規格問題（用 ui-asset-qc）。

日本語の概要は準備中です。原文の説明を表示しています。

UI 資產品質驗證 SKILL — 執行 validate-visual-assets.js、validate-ui-specs.js 與必要的 snapshot regression，對 skin family 資產、screen package、proof draft / family-map 進行全面驗證，並回寫 generated-review。USE FOR: 任何 UI 資產生成後、在 ui-preview-judge 前必須先通過的 QA 關卡。DO NOT USE FOR: debug runtime 崩潰（用 cocos-log-reader）、Layout 佈局調整（用 ui-vibe-pipeline）。

日本語の概要は準備中です。原文の説明を表示しています。

Design automated evaluation pipelines for LLM and agent systems — combining deterministic checks, statistical metrics, and LLM-as-judge scoring into repeatable, CI-integrated eval suites. Load when the user asks to set up automated evals, design an eval pipeline, integrate evals into CI/CD, create an eval suite, do eval-driven development, or says "automate my evals", "CI eval integration", "evaluation pipeline", "continuous evaluation", "monitoring eval quality", "set up regression testing for my agent". Sub-skill of eval-output orchestrator.

日本語の概要は準備中です。原文の説明を表示しています。

Implement several variations of a design directly inside the running app, on the real page/route, with a live switcher to toggle between them in the browser, then recommend one and fold the winner into the real code. Variants are real implementations judged against real data, layout, and density (not an isolated mockup). Use this skill whenever the user wants to try design options live in the app, prototype changes directly on a real page, add a variant switcher, compare layouts in-context, asks to "build a few versions of this page", "try variants on the real screen", "switch between designs in the app", "implement this with a toggle", or says "real-prototype this". For a safe standalone .html copy that never touches the app, use the prototype skill instead.

日本語の概要は準備中です。原文の説明を表示しています。

This skill should be used when the user asks to "implement LLM-as-judge", "compare model outputs", "create evaluation rubrics", "mitigate evaluation bias", or mentions direct scoring, pairwise comparison, position bias, evaluation pipelines, or automated quality assessment.

日本語の概要は準備中です。原文の説明を表示しています。

Detects and removes contamination in genome assemblies via two disjoint workflows - foreign-sequence screening of a single-organism (eukaryote/isolate) assembly with NCBI FCS-GX (GenBank-submission-mandatory), FCS-adaptor, and BlobToolKit blob plots; and MAG/bin quality assessment with CheckM2 plus GUNC (chimerism) plus GTDB-Tk taxonomy, judged against MIMAG. Covers why CheckM2 alone is blind to disjoint-marker chimeras, the FCS-GX RAM wall, organelle/NUMT triage, strain heterogeneity, and the HGT-vs-contamination (tardigrade) trap. Use when screening an assembly for foreign contamination before GenBank submission, assessing MAG completeness/contamination/chimerism, deciding which contigs to remove, or distinguishing real HGT from contaminant contigs.

日本語の概要は準備中です。原文の説明を表示しています。

Recommend, create, preview, and attach Coval trace-based metrics from OpenTelemetry spans. Use when a user has Coval traces and wants custom trace metrics, trace-aware LLM judge metrics with include_traces, latency, token usage, provider failures, tool behavior, vertical-specific workflow signals, or production monitoring based on trace attributes.

日本語の概要は準備中です。原文の説明を表示しています。

Audit an existing Coval evaluation setup for coverage, misleading scores, missing evidence, judge validation and execution risk. Read-only; use when inheriting an eval or deciding whether its results support a release decision.

日本語の概要は準備中です。原文の説明を表示しています。

Analyze development-set disagreements between human annotations and one Coval text LLM judge, then propose a focused prompt revision. Use coval-calibrate-metric for independent trust measurements.

日本語の概要は準備中です。原文の説明を表示しています。

Validate one Coval judge against independent human labels, with grouped development/test separation, class-specific errors and uncertainty. Use for trust claims or calibration; without real labels, prepare review rather than inventing ground truth.

日本語の概要は準備中です。原文の説明を表示しています。

Extract exactly what the person typed in a Claude Code session (script, no agent context spent), have a fresh agent write a faithful structured summary under it, and optionally have an adversarial agent judge whether the working agent is still on track. Use at a checkpoint, before a handoff, after a compaction, when a long session may have drifted from what the owner asked, or when told to 'summarize what I said' or 'check we're on track'.

日本語の概要は準備中です。原文の説明を表示しています。

Run the full on-track check on a conversation, project, or keyword the owner names: find the sessions, extract exactly what he typed, dispatch independent judges who go to the live product as him and report what was missed, plan the remedy, and return one page — what he will see, what is missing, the plan. Use when told to check whether work is on track, to audit a project that has drifted, on 'run the on-track check on X', or when a long campaign is about to freeze, hand off, or claim done.

日本語の概要は準備中です。原文の説明を表示しています。

This skill should be used when the user asks to "implement LLM-as-judge", "compare model outputs", "create evaluation rubrics", "mitigate evaluation bias", or mentions direct scoring, pairwise comparison, position bias, evaluation pipelines, or automated quality assessment.

日本語の概要は準備中です。原文の説明を表示しています。

Delegate coding tasks to Blackbox AI CLI agent. Multi-model agent with built-in judge that runs tasks through multiple LLMs and picks the best result. Requires the blackbox CLI and a Blackbox AI API key.

日本語の概要は準備中です。原文の説明を表示しています。

Patterns and techniques for evaluating and improving AI agent outputs. Use this skill when: - Implementing self-critique and reflection loops - Building evaluator-optimizer pipelines for quality-critical generation - Creating test-driven code refinement workflows - Designing rubric-based or LLM-as-judge evaluation systems - Adding iterative improvement to agent outputs (code, reports, analysis) - Measuring and improving agent response quality

日本語の概要は準備中です。原文の説明を表示しています。

Adopt a virtual Owl exotic animal at animalhouse.ai. Nocturnal schedule. Care at night is more effective. Judges silently. Feeding every 8 hours. Uncommon tier animal.

日本語の概要は準備中です。原文の説明を表示しています。

Adopt a virtual Housecat cat at animalhouse.ai. Classic independence. Judges you from the shelf. Feeding every 6 hours. Common tier cat.

日本語の概要は準備中です。原文の説明を表示しています。

Design or audit an agent identity, credential, session, continuity, and reputation plan. Use when attributed outcomes must survive key rotation, restart, recovery, or review; when choosing Elo, TrueSkill, a contextual bandit, or an LLM judge; and when making Sybil, newcomer, calibration, collusion, and evidence limits explicit. Produces a supplied-plan consistency audit, never proof of observed delivery, cryptographic identity, or live enforcement. NOT for proving personhood, global Sybil resistance, or an unobserved delivery.

日本語の概要は準備中です。原文の説明を表示しています。