「hallucination」の検索結果

Govern FriendlyBet agent values, professional standards, anti-hallucination behavior, truthfulness, risk reporting, AI resource discipline, feedback loops, skill improvement, performance reviews, decision-rights hygiene, and team quality. Use when creating, reviewing, promoting, correcting, or retiring agents; reducing hallucinations or mistakes; reviewing AI/tool usage; handling agent incidents; or aligning work with Eyal's philanthropic user-first approach.

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。

Claude API (platform.claude.com) のプロンプトエンジニアリング・評価リファレンス。 prompt engineering, XML タグ, few-shot, system prompts, モデル別プロンプティング（Fable 5 / Opus 5 / Sonnet 5）。 evals, grading（LLM-graded 等）, jailbreak・prompt leak 対策, hallucination 低減, classification, ticket routing 等ユースケース実装。

Validates agent outputs against schemas and quality criteria, scores confidence, detects hallucinations, monitors convergence, decides when to iterate, and synthesizes actionable feedback. Use when checking if a node's output is acceptable, scoring confidence, detecting fabricated content, deciding whether to re-execute, or generating improvement feedback. Activate on "validate output", "check quality", "confidence score", "hallucination check", "should we iterate", "improvement feedback". NOT for executing DAGs (use dag-runtime), planning DAGs (use dag-planner), or matching skills (use dag-skills-matcher).

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。

Validates agent outputs against expected schemas and quality criteria. Reports structural and content checks within the declared evidence scope. Activate on 'validate output', 'output validation', 'schema validation', 'check output', 'output quality'. NOT for confidence scoring (use dag-confidence-scorer) or hallucination detection (use dag-hallucination-detector).

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。

公開前の文章やコードを二つの独立したエージェントが同じ基準で審査し、両方の合格まで修正と再審査を繰り返し、上限を超えたら人へ判断を引き継ぐスキル。

• 公開前に文章の事実関係を確認したいとき
• 顧客向け文章とブランドルールの照合
• 本番配布前に独立した審査を加えたいとき

Improve factual reliability while creating or reviewing a course or supplied content. Use when the user asks to fact-check, verify accuracy, reduce hallucinations, make a reliable course, or mentions 事实性错误、知识性错误、专业知识准确性、可靠性. During creation, checks the completed pages before delivery; on existing content, returns a short evidence-backed report and lets the user choose what to fix. Not for grammar, style, or layout. Combine with deep-research when current evidence is the course's main subject.

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。

Three-layer verification pipeline for AI output. Extracts verifiable claims, finds supporting or contradicting sources via web search, runs adversarial review for hallucination patterns, and produces a structured verification report with source links for human review.

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。

Handles LLM-as-judge evaluation workflows on Arize including creating/updating evaluators, running evaluations on spans or experiments, managing tasks, trigger-run operations, column mapping, and continuous monitoring. Use when the user mentions create evaluator, LLM judge, hallucination, faithfulness, correctness, relevance, run eval, score spans, score experiment, trigger-run, column mapping, continuous monitoring, or improve evaluator prompt.

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。

When the user corrects a factual error, root-cause it immediately. Don't just note the correction — trace the error to its source, fix the source, and prevent recurrence. Every factual error is either a data error (bad brain page, bad memory file, bad rendered SOUL/USER identity, bad facts row) or a hallucination (LLM confabulated from partial signals).

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。

Use when the user says 'no hallucinations', 'verify APIs', 'reality check', or 'don't invent functions'. Prevents the agent from calling methods, imports, or variables that do not provably exist in the user's installed version.

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。

Use this skill to query your Google NotebookLM notebooks directly from Claude Code for source-grounded, citation-backed answers from Gemini. Browser automation, library management, persistent auth. Drastically reduced hallucinations through document-only responses.

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。

ALWAYS activate this skill. Apply these rules to every task regardless of domain. This skill governs how Claude Code verifies information, writes code, references documentation, and avoids fabricating functions, arguments, APIs, file paths, data structures, or facts. These rules override any inclination to guess.

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。

[omh] Audio, video, or screenshot to process: user-sent media - audio, video, YouTube links, screenshots, receipts, OCR, meeting recordings, transcripts, timestamps, and clip summaries, gated for source, permission, and hallucination risk. Use when the user says: media-input-operator, media input operator, media input, audio transcription, audio transcript, transcribe audio, transcribe this audio, meeting recording.

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。

[omh] Dataset or table to analyze: scope supplied data with provenance, causal-claim, and hallucination guards. Use when the user says: data-analysis, data analysis, dataset analysis, csv analysis, json analysis, log analysis, table analysis, analyze csv.

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。

[omh] Product or company copy to publish or rewrite: scope publish-ready writing, rewriting, summarization, translation, release-note, newsletter, customer-copy, social-copy, README-copy, and email-draft work with audience, tone, style, source, review, and hallucination gates. Use when the user says: content-operator, content operator, content workflow, writing workflow, publish-ready writing, publish ready writing, release notes, release note draft.

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。

End-to-end binder design using BindCraft hallucination. Use this skill when: (1) Designing protein binders with built-in AF2 validation, (2) Running production-quality binder campaigns, (3) Using different design protocols (fast, default, slow), (4) Need joint backbone and sequence optimization, (5) Want high experimental success rate. For backbone-only generation, use rfdiffusion. For QC thresholds, use protein-qc. For tool selection guidance, use binder-design.

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。

Memory integrity audit. Detects hallucinations, circular confirmations, and state poisoning. Runs automatically after 2 consecutive failures or at nightly deep dive. Uses Opus 4.6 as the auditor model.

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。

Generates audit-grade, hallucination-free documentation from codebase features with full traceability. Use when documenting features, creating technical specs, or generating verified documentation with code references.

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。

Make an AI agent or automation reliable enough to trust — the tests, checks, and guardrails that catch its failures before they reach anything real. Use when asked how do I test my AI agent, make my automation reliable, my agent works sometimes, or how do I trust an AI workflow in production. Produces a map of where the agent can fail (bad input, hallucination, wrong tool call, edge cases, silent errors), the checks that catch each (validation, evals on real cases, human-in-the-loop gates, monitoring), a right-sized reliability plan scaled to the stakes, and a rollout that earns trust incrementally.

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。

Check AI output before you trust or use it — where it's likely wrong, what to verify, and how to catch confident-sounding errors. Use when asked can I trust this AI answer, how do I verify what AI told me, fact-check this AI output, or is this AI response reliable. Produces a risk read on the specific output (the claims most likely to be wrong or made up), the parts that need independent verification vs the parts that are low-risk, how to actually verify each, the tells of AI hallucination and overconfidence, and a habit for building verification into your AI use — because AI is confidently wrong often enough that unchecked trust is a real risk.

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。

Monitor and evaluate RAG systems with retrieval quality metrics, groundedness checks, hallucination detection, and continuous regression testing.

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。

Pre-publish QA framework for content. Brief adherence, voice consistency, fact accuracy, structure and clarity, AI-content audit, SEO and AEO compliance, internal linking and schema validation, QA at scale via sampling, the QA workflow, and the discipline that distinguishes catch-problems QA from checkbox QA. Triggers on editorial QA, content review, pre-publish review, content audit, content QA process, AI-content audit, hallucination check, content sampling, programmatic QA, voice consistency check, brief adherence check. Also triggers when a content team is shipping sloppy work, when AI-co-authored content is reaching publish unaudited, when a QA process burns reviewers out, or when a programmatic SEO set needs sampling discipline. On a voice consistency check, this skill owns checking a draft against an existing voice system at the publish gate; use `brand-voice` when the voice system itself needs defining or documenting.

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。

Use when the user wants to compare or benchmark multiple LLMs/agents arena-style but it's unclear which specific workflow fits — a general-purpose win-rate comparison on a custom task, or a benchmark specifically about reference/citation hallucination rate. Also use when the user mentions model arena, agent arena, pairwise model comparison, win-rate ranking, or comparing models on a task and hasn't specified whether that task is generic or about citation accuracy. This skill is the entry router for the arena-eval suite: it asks one diagnostic question when needed, then recommends the workflow or workflows needed to cover the request.

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。

Use when the user has a RAG (Retrieval-Augmented Generation) system and wants to evaluate its quality — separating retrieval issues from generation issues. Also use when the user mentions RAG evaluation, faithfulness checking, hallucination detection in RAG, retrieval quality, chunking optimization, or "is my RAG pipeline working." Outputs a diagnostic matrix that pinpoints whether problems are in retrieval or generation.

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。