The Agent-Adoption Specification defines an open methodology for measuring how accessible a website is to AI agents and automated HTTP clients — a category that spans modern crawler-class clients used by LLM training, retrieval-augmented systems, and autonomous-agent architectures.
This repository hosts the canonical specification text, version history, and contribution process. The companion live human-readable page is published at respectarium.com/spec/agent-adoption/v1.
Current version: v1.0.0 (2026-04-26)
- Defines a list of measurable HTTP-level practices that determine whether a website is accessible to AI agents and automated HTTP clients
- Defines a scoring methodology that aggregates check results into a single 0–100 score and a three-level readiness ladder (integer levels 1, 2, 3)
- Defines a conformance contract — what an implementing scanner MUST and SHOULD do to be considered conformant with the Agent-Adoption Specification
- Bundles a B2B SaaS profile as the first implementation profile (other profiles for e-commerce, media, government, etc., may ship in future versions)
- Does NOT specify how individual LLMs rank brands (those algorithms are not publicly documented and not in scope)
- Does NOT make causal claims linking agent-readiness to LLM citation outcomes — the companion correlation research tests those claims empirically and finds modest, not dramatic, effects
- Does NOT mandate any specific implementation technology (you may use any HTTP client, programming language, or runtime to build a conforming scanner)
- Does NOT define UI/UX conventions for presenting scanner results to humans
- Does NOT impose pricing or commercial terms on any implementer
This specification is open. Anyone may implement a scanner against it. To date, the following implementations exist:
- Respectarium agent-adoption Check tool (proprietary commercial implementation; the version used to produce the 2026-04 correlation study is documented in the study release notes)
We invite additional independent implementations. Cross-implementation divergence on shared check definitions is itself a research-relevant phenomenon — see the correlation study's findings on cross-scanner agreement for empirical context.
The canonical specification text lives in SPEC.md. The same text is rendered as a human-readable web page at respectarium.com/spec/agent-adoption/v1.
The two views are kept in sync — edits land in this repository first via Pull Request, then the live page is updated. When in doubt, the SPEC.md in the latest tagged release is authoritative.
The specification is implementation-agnostic. To build a conforming scanner:
- Read
SPEC.md, particularly the per-check definitions and conformance section - Implement an HTTP client that exercises the checks against target domains
- Apply the scoring formula and level-gate logic as specified
- Produce output conforming to the canonical JSON schema at
schemas/output-v1.schema.json(described inSPEC.md§10) - Optionally, run your scanner against the 908 brand domains in the published study dataset and compare your per-check results to the published reference outputs (cross-scanner divergence is itself a documented finding — see the correlation study)
We welcome contributions — methodology refinements, new check proposals, conformance clarifications, and corrections.
- For non-developers: open an Issue using one of the templates
- For developers: open a Pull Request against
mainwith your proposed changes; see CONTRIBUTING.md for guidelines
The decision process for accepting changes is in governance/maintainers.md.
- v1.0.x — current line; editorial fixes, clarifications without changing behavior
- v1.1.0 — additive changes (new optional checks, new profiles)
- v2.0.0 — breaking changes (renamed checks, score-formula changes, removed checks, new gate criteria)
Each tagged release is immutable once published. Errata for past tags are documented in CHANGELOG.md with pointers to the corrected analysis in subsequent tags.
The next major version (v2.0) is planned for ~Q3 2026, calibrated against findings from the quarterly Agent-Adoption Correlation Study. v2 will recalibrate check weights based on empirical signal-to-outcome relationships established in those studies.
This specification is licensed under the Creative Commons Attribution 4.0 International License (CC-BY 4.0).
You may copy, distribute, modify, and build upon this work for any purpose, including commercial, provided you give appropriate credit to Respectarium.
Suggested attribution:
Source: Agent-Adoption Specification by Respectarium — https://respectarium.com/spec/agent-adoption/v1
Full license text: LICENSE
When citing this specification in research, implementations, or derived work:
Agent-Adoption Specification, Version 1.0. Respectarium, 2026-04-26. Available at: https://respectarium.com/spec/agent-adoption/v1 Source: https://github.com/respectarium/agent-adoption-spec/releases/tag/v1.0.0
This specification is part of a broader research and methodology program at Respectarium:
- Agent-Adoption Correlation Study — quarterly empirical research testing which signals defined in this specification correlate with LLM visibility outcomes. The Q1 study (
study-2026-04) is the foundation for evidence-based v2 design. respectarium.com/research/correlation-2026-04— the formal preprint of the Q1 correlation study- Respectarium's hosted implementation — a hosted scanner conforming to this specification; available at respectarium.com
Maintained by Respectarium. See governance/maintainers.md for the decision process and contact information.
For private inquiries: open a Discussion on this repository.