Agents Inherit Your Governance — Whether You Meant Them To or Not
FOST Paris · CNIT Forest, La Défense · 1–3 December 2026
Kin Lane — API Evangelist
PLACEHOLDER — rewrite for the room
You are not choosing whether agents use your APIs
They are already calling them. Whatever governance you have
— or do not have — is what those agents inherited the day they arrived.
Fourth outing of this spine, first in Paris.
PLACEHOLDER — rewrite for the room
"AI-ready"
Claimed by every API program, measured by almost none. So I
measured it — across the whole APIs.io catalog (26,755 providers), on
two instruments that never touch each other.
Kin Score 0.19.0
The Two Instruments
Kin Score
0–100
composite of governance fundamentals
6 base facets
+ 2 conditional —
regulatory · open source
never blended
Agent Readiness
0–100
standalone — can an agent drive it safely?
19 dimensions
139 raw points
Never blended Computed independently — which is what makes
the relationship between them a finding rather than an assumption.
The finding — measured, 2026-09-06
Readiness is a layer. A layer does not rise above its foundation.
r = 0.882composite × agent readiness, 26,755 providers
3.3%score agent readiness ABOVE their composite
Composite band
Providers
Mean agent readiness
Reaching Agent-Native
Exemplar
273
42.6
61.5%
Strong
1,164
35.6
30.9%
Developing
3,912
28.3
10.6%
Thin
4,412
19.7
2.0%
Emerging
6,551
4.3
0.2%
Minimal
10,443
0.2
0.0%
Part one
The Composite, Facet by Facet
6 base facets, 2 conditional.
Composite facet 1 of 8
Contract Quality
Measures how deep and machine-readable the API contract itself is — OpenAPI, AsyncAPI, JSON Schema, JSON-LD.
0.25 of the composite
18.0mean
0.0median
63.6%score zero
22.2%50 or above
When the consumer is an agent A human reads the docs when the contract is thin; an agent has nothing else to read.
Composite facet 2 of 8
Developer Ergonomics
Measures how quickly someone can get from landing page to a working call — SDKs, CLI, portal, getting-started, sandbox, auth clarity.
0.20 of the composite
21.2mean
9.5median
21.9%score zero
16.0%50 or above
When the consumer is an agent Every affordance here was built to shorten a human's first hour, and an agent has no first hour — it either works on the first call or it does not.
Composite facet 3 of 8
Access Clarity
Measures whether what it costs, what you may do with it, and how you get in are all published — plans, pricing, sign-up, terms, privacy.
0.20 of the composite
24.1mean
21.1median
29.4%score zero
13.6%50 or above
When the consumer is an agent An agent cannot accept terms it cannot read, and cannot sign up for a key through a form.
Composite facet 4 of 8
Operational Transparency
Measures whether the provider publishes how the API BEHAVES in production — rate limits, status, change log, deprecation, security disclosure.
0.13 of the composite
12.6mean
2.6median
44.3%score zero
5.9%50 or above
When the consumer is an agent Rate limits, status and deprecation are the signals a human absorbs by noticing; an agent only knows what is published as data.
Composite facet 5 of 8
Contract Governance
Measures the artifacts that describe and constrain the contract — rulesets, vocabulary, declared conformance, overlays.
0.12 of the composite
5.8mean
0.0median
64.8%score zero
0.5%50 or above
When the consumer is an agent Rulesets and overlays are how a provider says what its contract PROMISES, which is the only thing an agent can plan against.
Composite facet 6 of 8
Discoverability
Measures whether the API can be found and understood from machine-readable metadata alone, with no human reading the site.
0.10 of the composite
59.5mean
57.4median
0.0%score zero
92.9%50 or above
When the consumer is an agent An agent that cannot find the contract never gets to the part where the contract is good.
Composite facet 7 of 8
Regulatory Posture
Measures, for regulated providers only, whether the API publishes the consent, security, legal and conformance posture its regime demands.
0.15 of the composite · conditional
Applies to 9,442 providers (35.3% of the catalog)
26.4mean
20.4median
4.7%score zero
14.8%50 or above
When the consumer is an agent A regime tells you which failures are not recoverable, which is exactly what an agent needs to know before it acts.
Composite facet 8 of 8
Open Source Surface
Measures, where the product is itself an open repository, whether it publishes the maintainership surface you need to depend on it.
0.10 of the composite · conditional
Applies to 681 providers (2.5% of the catalog)
57.7mean
60.0median
7.0%score zero
64.5%50 or above
When the consumer is an agent If the product is the repository, the maintainership surface IS the operational contract.
Part two
Agent Readiness, Dimension by Dimension
19 dimensions · 139 raw points ·
scored independently of everything in part one.
Agent Readiness dimension 1 of 19
Machine-Readable Contract
Is there a public OpenAPI contract at all?
18 of 139 raw points
9,089 of 26,755 providers earn it
(34.0%)
present9,089 · 34.0%
absent17,666 · 66.0%
When the consumer is an agent With no machine-readable contract there is no programmatic surface to drive at all.
Agent Readiness dimension 2 of 19
MCP Server
Is there a Model Context Protocol server an agent can actually connect to?
12 of 139 raw points gradedprovenance-graded
2,288 of 26,755 providers earn it
(8.6%)
verified828 · 3.1%
documented only1,460 · 5.5%
absent24,467 · 91.4%
When the consumer is an agent The most direct expression of agent-first design, and the easiest to claim without shipping.
Agent Readiness dimension 3 of 19
Agentic Access Contract
Is every operation classified by action class, consequence, and when a human must be asked?
10 of 139 raw points provenance-graded
6,527 of 26,755 providers earn it
(24.4%)
present38 · 0.1%
DERIVED — written by API Evangelist, not the provider6,489 · 24.3%
absent20,228 · 75.6%
When the consumer is an agent The difference between an agent knowing an operation exists and knowing what it costs to be wrong about it.
Agent Readiness dimension 4 of 19
Machine-Readable Auth
Can an agent authenticate with a credential bound to it, or only bear a static secret?
10 of 139 raw points graded
10,201 of 26,755 providers earn it
(38.1%)
served at the provider's own host1,554 · 5.8%
bound credential14 · 0.1%
negotiable OAuth740 · 2.8%
static bearer secret7,893 · 29.5%
absent16,554 · 61.9%
When the consumer is an agent An agent either holds a credential bound to it, or it bears a secret that names nobody.
Agent Readiness dimension 5 of 19
Idempotency
Do mutating operations accept an idempotency key, so a retry is not a second write?
9 of 139 raw points graded
898 of 26,755 providers earn it
(3.4%)
verified319 · 1.2%
documented only579 · 2.2%
not applicable (no write surface)1,076 · 4.0%
absent24,781 · 92.6%
When the consumer is an agent Agents retry. Without an idempotency key a retry is a second charge, a duplicate ticket, a message sent twice.
Agent Readiness dimension 6 of 19
Stable Error Semantics
Are errors a stable, typed envelope an agent can branch on?
8 of 139 raw points graded
5,503 of 26,755 providers earn it
(20.6%)
verified2,415 · 9.0%
documented only3,088 · 11.5%
absent21,252 · 79.4%
When the consumer is an agent Agents branch on errors; free text sends them into blind retries.
Agent Readiness dimension 7 of 19
A2A Agent Card
Does the provider serve an A2A agent card at its own well-known path?
8 of 139 raw points graded
99 of 26,755 providers earn it
(0.4%)
conformant28 · 0.1%
near-conformant18 · 0.1%
shaped like one, fails a hard check53 · 0.2%
absent26,656 · 99.6%
When the consumer is an agent The only agent surface in this rubric that cannot be derived on a provider's behalf — it is served from their own host or it does not exist.
Agent Readiness dimension 8 of 19
Request/Response Examples
Do operations carry request and response examples an agent can learn the payload shape from?
7 of 139 raw points graded
4,304 of 26,755 providers earn it
(16.1%)
verified1,875 · 7.0%
partial coverage1,754 · 6.6%
documented only675 · 2.5%
absent22,451 · 83.9%
When the consumer is an agent The cheapest way an agent learns a payload shape before it ever makes a call.
Agent Readiness dimension 9 of 19
Rate-Limit Signaling
Is rate-limit state published as machine-readable data, ideally in response headers?
7 of 139 raw points graded
9,578 of 26,755 providers earn it
(35.8%)
verified369 · 1.4%
documented only9,209 · 34.4%
absent17,177 · 64.2%
When the consumer is an agent Documented limits tell an agent the ceiling; response headers tell it where it currently is.
Agent Readiness dimension 10 of 19
Documented Reversibility
Does the contract say what can be undone, how, and within what window?
6 of 139 raw points graded
1,878 of 26,755 providers earn it
(7.0%)
verified187 · 0.7%
documented only1,691 · 6.3%
not applicable (no write surface)1,076 · 4.0%
absent23,801 · 89.0%
When the consumer is an agent Dry-run lets an agent rehearse and idempotency stops it double-firing; neither tells it whether the thing it just did can be taken back.
Agent Readiness dimension 11 of 19
Typed Event Surface
Is the webhook or event surface described by a contract rather than an example payload?
6 of 139 raw points
2,828 of 26,755 providers earn it
(10.6%)
present2,090 · 7.8%
DERIVED — written by API Evangelist, not the provider738 · 2.8%
absent23,927 · 89.4%
When the consumer is an agent An agent reacting to state changes needs a typed surface, not an example payload in a blog post.
Agent Readiness dimension 12 of 19
Delegated User Identity
Can a caller obtain a token scoped to the human it is acting for?
6 of 139 raw points graded
2,039 of 26,755 providers earn it
(7.6%)
served at the provider's own host1,384 · 5.2%
documented only655 · 2.4%
absent24,716 · 92.4%
When the consumer is an agent Can a caller get a token scoped to the HUMAN it is acting for, or only a credential belonging to the integration?
Agent Readiness dimension 13 of 19
Registration Without a Human
Can a new client register itself without a developer filling in a portal form?
6 of 139 raw points
817 of 26,755 providers earn it
(3.1%)
present817 · 3.1%
absent25,938 · 96.9%
When the consumer is an agent If every new agent needs a developer to fill in a portal form and paste a key, there is no agent onboarding path.
Agent Readiness dimension 14 of 19
Agent Skills
Are there packaged operating instructions for an agent using the API?
5 of 139 raw points provenance-graded
4,102 of 26,755 providers earn it
(15.3%)
present747 · 2.8%
DERIVED — written by API Evangelist, not the provider3,355 · 12.5%
absent22,653 · 84.7%
When the consumer is an agent Packaged operating instructions, rather than leaving usage to be inferred from the spec.
Agent Readiness dimension 15 of 19
Protected Resource Metadata
Can an agent discover at runtime how a resource is protected and by which authorization server?
5 of 139 raw points graded
598 of 26,755 providers earn it
(2.2%)
verified597 · 2.2%
documented only1 · 0.0%
absent26,157 · 97.8%
When the consumer is an agent An agent discovers how a resource is protected and which authorization server protects it, at runtime, instead of being configured by a human.
Agent Readiness dimension 16 of 19
Agentic Commerce Well-Known Document
Is there a conformant agentic-commerce document, and is the surface the provider’s own?
5 of 139 raw points graded
229 of 26,755 providers earn it
(0.9%)
the provider's own surface5 · 0.0%
a vendor's surface, under their name224 · 0.8%
absent26,526 · 99.1%
When the consumer is an agent Whether an agent can transact here — and whose surface it is transacting on.
Agent Readiness dimension 17 of 19
Well-Known Catalog
Is there a /.well-known/api-catalog linkset an agent can hit first?
4 of 139 raw points
484 of 26,755 providers earn it
(1.8%)
present484 · 1.8%
absent26,271 · 98.2%
When the consumer is an agent The canonical machine entry point an agent hits first.
Agent Readiness dimension 18 of 19
Dry-Run / Simulate Mode
Can a destructive operation be planned and previewed before it is committed?
4 of 139 raw points
125 of 26,755 providers earn it
(0.5%)
present125 · 0.5%
not applicable (no write surface)1,076 · 4.0%
absent25,554 · 95.5%
When the consumer is an agent Every other safety rail here limits the damage of an action already taken; this is the only one that prevents it.
Agent Readiness dimension 19 of 19
Consent & Bot Identity
Are AI usage preferences and agent traffic identity published in machine-readable form?
3 of 139 raw points
85 of 26,755 providers earn it
(0.3%)
present85 · 0.3%
absent26,670 · 99.7%
When the consumer is an agent Machine-readable AI usage preferences, and cryptographically identified agent traffic.
The band gate
This is not the future I would have chosen, but it is the one we have.
The market responded the way it did to LLMs and agentic automation.
It may not be the ideal approach to take, but it is the market we have, no matter
how irrational.
Lean In I've chosen to lean into the market rather than
fight it, making my approach an offensive and defensive approach to governing how
agents engage with my business.
PLACEHOLDER — rewrite for the room
Where do you stand with your API?
Look yourself up before you leave the room.
If the catalog is wrong about you, tell me — corrections are
free, and I will rescore you.
