Methodology & Bibliography
How the Agentic Glossary is built, what counts as a canonical citation, and how we keep the entries fresh.
Data last refreshed: 2026-05-07. Next scheduled refresh: Q3 2026 (or sooner on regulator / vendor trigger).
Sourcing rule
Every glossary entry has at least one direct quote from a canonical primary source — meaning:
- The vendor's own published documentation, blog, or research paper (Anthropic, Google, OpenAI, Microsoft, Meta).
- The standards body's own specification (Linux Foundation A2A / MCP, NIST, ISO).
- The regulator's own publication (European Commission, NIST, FATF, FCA).
- The peer-reviewed paper of record (arXiv-published papers cited >500 times, NeurIPS / ICLR proceedings).
- An analyst publication where the analyst is the primary source for the framing (Gartner Hype Cycle terminology, McKinsey adoption stats).
Wikipedia, secondary blogs, and content farms are never the primary citation — they may appear as supplementary reading on related entries.
Freshness flags
Every entry carries a freshness flag:
- Foundational — citations to original peer-reviewed papers (e.g., the 2020 RAG paper, the 2022 Constitutional AI paper). Considered evergreen — refreshed only if substantially revised.
- In force — current, binding regulator text or in-effect specifications. Includes the in-force date (e.g., EU AI Act Article 6, 2 August 2026).
- Emerging 2026 — terms that entered mainstream discourse in 2026 (ADLC, AX, context graph, agent management platform). Quarterly review.
- Contested — entries where there is meaningful, named disagreement in the field. We carry both positions.
Anything older than six months that does not carry one of the four flags is considered stale and gets retired or refreshed.
Refresh cadence
| Trigger | Action |
|---|---|
| Quarterly | Full audit: every cited URL pinged, every primary source re-read for material changes |
| Vendor canonical-source publication (Anthropic, Google, OpenAI, NIST, etc.) | Targeted refresh of affected entries within 7 days |
| Regulator publication or in-force date change | Same-week update |
| New peer-reviewed paper that supersedes a cited claim | Same-week update |
| URL 404 or vendor pivot | Immediate fix |
Bibliography (v0 — 2026-05-07)
The full canonical-source list backing the v0 entry set. Each line: source, URL, accessed date, role.
- Anthropic, Building Effective Agents. www.anthropic.com/research/building-effective-agents
- Anthropic, Measuring AI agent autonomy in practice. www.anthropic.com/research/measuring-agent-autonomy
- Bai et al. (Anthropic), Constitutional AI: Harmlessness from AI Feedback, arXiv:2212.08073, 2022. arxiv.org/abs/2212.08073
- Lewis et al., Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks, arXiv:2005.11401, 2020. arxiv.org/abs/2005.11401
- Anthropic, Introducing the Model Context Protocol. www.anthropic.com/news/model-context-protocol
- Model Context Protocol, Specification (2025-11-25 revision). modelcontextprotocol.io/specification/2025-11-25
- Google Developers, Announcing the Agent2Agent Protocol (A2A). developers.googleblog.com/en/a2a-a-new-era-of-agent-interoperability/
- A2A Protocol, Specification. a2a-protocol.org/latest/specification/
- NIST, Artificial Intelligence Risk Management Framework (AI 100-1). nvlpubs.nist.gov/nistpubs/ai/nist.ai.100-1.pdf
- NIST, AI Risk Management Framework hub (program announcements; AI Agent Standards Initiative announced February 2026 via CAISI). nist.gov/itl/ai-risk-management-framework
- European Commission, EU AI Act Implementation Timeline. ai-act-service-desk.ec.europa.eu/en/ai-act/timeline/timeline-implementation-eu-ai-act
- Gartner, Hype Cycle for Agentic AI 2026. gartner.com/en/articles/hype-cycle-for-agentic-ai
- McKinsey QuantumBlack, The state of AI in 2025: Agents, innovation, and transformation, November 2025. mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai
- Marcus, G., Six (or seven) predictions for AI 2026 from a Generative AI realist, Substack. garymarcus.substack.com/p/six-or-seven-predictions-for-ai-2026
- AgentsBooks, Anatomy of a Firm. agentsbooks.com/anatomy
What counts as a "term"?
The bar for inclusion in v0:
- It is searched for, currently, by builders / operators / buyers of AI agents (validated via search-intent signals — Reddit, HN, Google Trends).
- At least one canonical primary source defines it directly.
- It connects to one of the eight AgentsBooks primitives or to a current pillar topic (P1–P8).
v0 ships with 31 terms across six categories. v1 will expand toward 100+ as new vocabulary stabilises.
How to suggest a term or correction
Email hello@agentsbooks.com with the term, what you'd like it defined as, and the canonical source you want quoted. We aim to triage within five business days; verified additions ship in the next quarterly refresh, with same-week handling for regulator / vendor news.