Cognition · Data artifact

Prompt Exemplar Set

Data artifactCognitionCognition & Memoryarc:PromptExemplarSet

A curated collection of task-specific example prompts and reasoning traces supplied to the model to steer agent reasoning and tool selection.

Responsibility. Steers agent reasoning through task-matched examples.

Also known as: Few-shot exemplars, Instance-specific examples, Demonstration pool, Few-shot prompt pool, Chain-of-thought exemplars, Contrastive demonstrations, Few-shot examples, Judge calibration examples, CoT demonstration pool

Direct neighbourhood (hover for relationship types)

Relationships

configures structural

is read by dependency

is written by dependency

is evaluated by assurance

LLM Judge abstract Ch3.5

is produced by lifecycle

Prompt Optimizer Ref8.05

Design guidance

SHOULD closely match exemplars to the query task, since ReAct gains depend mainly on exemplar-query similarity.
MUST contain correctly labeled examples; quality matters more than quantity.
SHOULD be representative of the production input distribution and diverse enough to cover edge cases and difficult variants.
SHOULD be checked for balanced label and scenario distribution across critical dimensions.
MAY include correct and incorrect reasoning or routing examples to teach what to avoid.
SHOULD include examples covering the full range of tool output values the agent must interpret.
SHOULD curate the most informative few-shot examples, reduce their count and compress them to balance accuracy against token consumption.
SHOULD provide LLM judges with 5-10 annotated examples of hallucinated, partially grounded and fully grounded outputs.
SHOULD include only the most informative examples, since examples consume context capacity and longer contexts add latency (Ref5.06).

Quantitative guidance

As stated by the sources; verify before use.

Few-shot prompting improved accuracy by 28.2% over zero-shot on some tasks (Ch3.5).
Mislabeled demonstrations degraded intent-classification accuracy by 15-20 points (Ch3.5).
Five high-quality diverse examples outperform fifty low-quality or redundant ones (Ch3.5).
Eight chain-of-thought exemplars gave a 540B-parameter model state-of-the-art math word-problem results (Ch3.5).
Ablating few-shot examples degrades performance 5-15%, more for complex tasks and smaller models (Ch3.7).
Few-shot gains: 1-3 examples give dramatic improvement, 5-10 near-optimal, 20+ diminishing returns (Ref5.06).
Removing unnecessary examples saves 10-15% of tokens; a 1,200-token prompt with 12 examples was reduced to 400 tokens (Ref8.05).

Classification

Patterns: Few-shot promptingFew-shot in-context learningChain-of-thought promptingContrastive chain-of-thoughtFaithful chain-of-thought (reasoning plus code)
Quality attributes: Functional suitability: correctness and validity (ISO/IEC 25010 | NIST AI RMF: valid)Flexibility (ISO/IEC 25010)
Risks mitigated: Prompt sensitivityMislabeled demonstrations teaching wrong decision boundariesBiased example sets transmitting biasMulti-agent miscoordination

Sources

Ch1.2: T. Nguyen, "Core Agent Patterns," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 1.2. ISBN: 9798244538229.
Ch3.5: T. Nguyen, "Prompt Optimization, Few-Shot Learning, Fine-Tuning," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 3.5. ISBN: 9798244538229.
Ch3.6: T. Nguyen, "Trace Analysis and Execution Debugging," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 3.6. ISBN: 9798244538229.
Ch3.7: T. Nguyen, "Tool Usage Auditing," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 3.7. ISBN: 9798244538229.
Ch3.10: T. Nguyen, "Efficiency Metrics," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 3.10. ISBN: 9798244538229.
Ch5.1: T. Nguyen, "Chain-of-Thought (CoT) Fundamentals," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 5.1. ISBN: 9798244538229.
Ch5.9: T. Nguyen, "Working Memory," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 5.9. ISBN: 9798244538229.
Ch8.2A: T. Nguyen, "Error Rates and Reliability," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 8.2A. ISBN: 9798244538229.
Ch8.3: T. Nguyen, "Token Economics and Architecture," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 8.3. ISBN: 9798244538229.
Ref5.06: "Large Language Models Are In-Context Learners," unpublished reference note (06-LLM-In-Context-Learning-Research.md), Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam supplementary materials, 2026. unpublished note
Ref8.05: "Cost Optimization and Resource Monitoring for Agent Systems," unpublished reference note (05-Cost-Optimization-Resource-Monitoring.md), Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam supplementary materials, 2026. unpublished note