vendor-first-partyBuilding effective agents
Vendor first-party engineering guidance.
No diagram or asset reuse.
Evidence register
Inspect the source register.
Review every official, primary, community, standard, and vendor first-party source admitted to the launch registry—along with its scope and rights boundary.
Inspected launch sources
Evidence with limits attached.
Publication and review dates identify the registry record. They are not a promise that a source will never change.
External source links open in a new tab.
- vendor-first-partyDemystifying evals for AI agents
Vendor first-party practice; no universal or LucyAI-result claim.
No diagram or asset reuse.
- government-guidanceNIST AI 600-1: Generative Artificial Intelligence Profile
Voluntary government guidance; not law, certification, or proof of conformance.
Attribute and check embedded third-party material before reuse.
- primary-researchRetrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
Historical, model-specific, and dataset-specific primary research.
Cite and paraphrase; do not reuse figures or tables.
- primary-researchLost in the Middle: How Language Models Use Long Contexts
Results apply to the tested models and tasks, not every current model.
Attribution and change notice required for reuse.
- vendor-first-partyIntroducing Contextual Retrieval
Vendor first-party experiment; reported results remain Anthropic's scoped results.
No diagram or asset reuse.
- government-guidanceNIST SP 800-218A: Secure Software Development Practices for Generative AI
Secure-development guidance; not proof of security or conformance.
Attribute and check embedded third-party material before reuse.
- government-guidanceNIST AI 100-2 E2025: Adversarial Machine Learning
Government taxonomy and guidance; no security guarantee.
Attribute and check errata before detailed quotation.
- community-guidanceOWASP Top 10 for Agentic Applications 2026
Community prioritization framework, not certification or a complete security program.
Attribute; share adaptations under the applicable license.
- standardISO/IEC/IEEE 42010:2022
The inspected abstract does not support a conformance claim.
Do not reproduce clauses, tables, or paywalled text.