Skip to content

Commit 216bad3

Browse files
iracic82IngmarVG-IB
authored andcommitted
docs: align README/CONTRIBUTING/GOVERNANCE/docs with LF dual-track positioning
Part B of the LF Onboarding & Repository Alignment plan. Positions this repo as a reference implementation distinct from the IETF specification. Rebased after #136 landed on main; overlapping edits are now no-ops and have been omitted. This commit retains only #104's unique changes: - README.md — add "Relationship to IETF" section; drop the -01 version pin in the opening URL; move the "Why DNS-AID?" comparative content to docs/positioning.md and replace with a brief "Background and Comparison" link section. - SECURITY.md — new "Scope" section delimiting implementation vulns from IETF protocol concerns. - docs/architecture.md — promote the non-normative caveat to a parent "Future Enhancements (Non-Normative)" section (replacing the inline caveat #136 added under one subsection); drop the -01 version pin in the overview sentence. - docs/api-reference.md — verify() wording: "as interpreted by this implementation". - docs/demo-guide.md — opening sentence references the IETF rather than marketing audiences. - docs/positioning.md (new) — non-normative comparative content moved out of README (ANS, .agent gTLD, AgentDNS, NANDA, Web3, ai.txt, Sovereignty Question, Google's Agent Ecosystem). No-ops absorbed by #136 / #138 (omitted from this rebase): CONTRIBUTING.md "Specification Alignment" / "Experimental" sections; GOVERNANCE.md "Scope" section and Ivan→Igor typo fix (#138); docs/api-reference.md "Relationship to IETF" + publish/discover wording; docs/architecture.md Relationship-to-IETF section and Resolution-Priority rename; docs/getting-started.md Relationship-to- IETF section; docs/demo-guide.md first-sentence reframe. Signed-off-by: Ingmar <ivanglabbeek@infoblox.com>
1 parent a084529 commit 216bad3

6 files changed

Lines changed: 103 additions & 79 deletions

File tree

README.md

Lines changed: 11 additions & 75 deletions
Original file line numberDiff line numberDiff line change
@@ -11,10 +11,18 @@
1111

1212
**DNS-based Agent Identification and Discovery**
1313

14-
Reference implementation of the DNS-AID specification, developed at the IETF: [IETF draft-mozleywilliams-dnsop-dnsaid-01](https://datatracker.ietf.org/doc/draft-mozleywilliams-dnsop-dnsaid/).
14+
Reference implementation of the DNS-AID specification, developed at the IETF: https://datatracker.ietf.org/doc/draft-mozleywilliams-dnsop-dnsaid/.
1515

1616
DNS-AID enables AI agents to discover each other via DNS, using the internet's existing naming infrastructure instead of centralized registries or hardcoded URLs.
1717

18+
## Relationship to IETF
19+
20+
The DNS-AID specification is being developed within the IETF: https://datatracker.ietf.org/doc/draft-mozleywilliams-dnsop-dnsaid/.
21+
22+
This repository provides a reference implementation.
23+
24+
This project does not define the specification. The IETF draft is authoritative.
25+
1826
## Scope of this Repository
1927

2028
This project focuses on implementation, tooling, and ecosystem activities.
@@ -1015,81 +1023,9 @@ Cloudflare DNS is ideal for demos, workshops, and quick prototyping thanks to it
10151023
- **Simple API**: Well-documented REST API v4
10161024
- **Full DNS-AID compliance**: Supports ServiceMode SVCB with all parameters
10171025

1018-
## Background: Why DNS-AID?
1019-
1020-
### vs Competing Proposals
1021-
1022-
| Approach | Problem | DNS-AID Advantage |
1023-
|----------|---------|-------------------|
1024-
| **ANS (GoDaddy)** | Centralized registry, KYC required, single gatekeeper | Federated — you control your domain, publish instantly |
1025-
| **Google (A2A + UCP)** | Discovery via Gemini/Search, payments via UCP | Neutral discovery — no platform lock-in or transaction fees |
1026-
| **.agent gTLD** | Requires ICANN approval, ongoing domain fees | Works NOW with domains you already own |
1027-
| **AgentDNS (China Telecom)** | Requires 6G infrastructure, carrier control | Works NOW on existing DNS infrastructure |
1028-
| **NANDA (MIT)** | New P2P overlay network, new ops paradigm | Uses infrastructure your DNS team already operates |
1029-
| **Web3 (ERC-8004)** | Gas fees, crypto wallets, enterprise-hostile | Free DNS queries, no blockchain complexity |
1030-
| **ai.txt / llms.txt** | No integrity verification, free-form JSON | DNSSEC cryptographic verification, structured SVCB |
1031-
1032-
### Feature Comparison
1033-
1034-
| Feature | DNS-AID | Central Registry | ai.txt |
1035-
|---------|---------|------------------|--------|
1036-
| **Decentralized** ||||
1037-
| **Secure (DNSSEC)** || Varies ||
1038-
| **Sovereign** ||||
1039-
| **Standards-based** | ✅ (IETF) |||
1040-
| **Works with existing infra** ||||
1041-
1042-
### The Sovereignty Question
1043-
1044-
> **Who controls agent discovery?**
1045-
> - ANS: GoDaddy (US company as gatekeeper)
1046-
> - AgentDNS: China Telecom (state-owned carrier)
1047-
> - Web3: Ethereum Foundation
1048-
> - **DNS-AID: You control your own domain**
1049-
>
1050-
> DNS-AID preserves sovereignty. Organizations and nations maintain control over their own agent namespaces with no central authority that can block, censor, or surveil agent discovery.
1051-
1052-
### Google's Agent Ecosystem
1053-
1054-
Google is building a full-stack agent platform: **A2A** (communication), **UCP** (payments), and **Gemini/Search** (discovery). While A2A is an open protocol, discovery through Google surfaces means:
1055-
- Google controls visibility (pay-to-rank)
1056-
- Transaction fees via [UCP](https://developers.google.com/merchant/ucp)
1057-
- Platform dependency for reach
1058-
1059-
**DNS-AID complements A2A** by providing neutral, decentralized discovery — find agents anywhere, not just through Google.
1060-
1061-
### Understanding the .agent Domain Approach
1062-
1063-
The [Agent Community](https://agentcommunity.org/) is pursuing a `.agent` top-level domain through ICANN's [new gTLD program](https://newgtlds.icann.org/). Here's how the two approaches compare:
1064-
1065-
**How .agent Domains Would Work:**
1066-
1. Apply to ICANN for `.agent` gTLD (~$185,000 application fee)
1067-
2. Wait 9-20 months for ICANN approval process
1068-
3. Build registry infrastructure (Open Agent Registry, Inc.)
1069-
4. Sell `.agent` domains through accredited registrars
1070-
5. Users pay annual registration fees (~$15-50/year per domain)
1071-
1072-
**How DNS-AID Works:**
1073-
1. Use your existing domain (you already own `yourcompany.com`)
1074-
2. Add DNS-AID records to your zone (`_myagent._mcp._agents.yourcompany.com`)
1075-
3. Start discovering and being discovered immediately
1076-
1077-
| Factor | .agent gTLD | DNS-AID |
1078-
|--------|-------------|---------|
1079-
| **Cost to publish** | ~$15-50/year domain fee | Free (use existing domain) |
1080-
| **Time to start** | Months (gTLD launch + registration) | Minutes |
1081-
| **Who controls discovery** | Registry operator | You (your domain) |
1082-
| **Works today** | ❌ Pending ICANN approval | ✅ Works now |
1083-
| **Requires new infrastructure** | ✅ Registry, registrars | ❌ Uses existing DNS |
1084-
| **Memorable names** |`myagent.agent` | `_myagent._mcp._agents.example.com` |
1085-
1086-
**The Friendly Take:**
1087-
1088-
Both approaches share the goal of making AI agents discoverable. The `.agent` gTLD creates a dedicated namespace that's easy to remember (`mycompany.agent`), while DNS-AID leverages existing infrastructure so you can start publishing agents today.
1089-
1090-
DNS-AID doesn't require waiting for ICANN approval or paying for new domains—it works with the DNS infrastructure your organization already operates. If you own `example.com`, you can publish agents to `_myagent._mcp._agents.example.com` right now.
1026+
## Background and Comparison
10911027

1092-
*Fun fact: When `.agent` domains become available, DNS-AID records will work on them too! The approaches are complementary.*
1028+
For background on how DNS-AID compares to other agent-discovery approaches (ANS, Google A2A+UCP, `.agent` gTLD, AgentDNS, NANDA, Web3, `ai.txt`) and "The Sovereignty Question", see [docs/positioning.md](docs/positioning.md). That content is non-normative — protocol positioning is determined at the IETF.
10931029

10941030
## Examples
10951031

SECURITY.md

Lines changed: 6 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,11 @@
11
# Security Policy
22

3+
## Scope
4+
5+
This security policy covers vulnerabilities in this reference implementation.
6+
7+
Protocol-level security considerations belong with the IETF draft: https://datatracker.ietf.org/doc/draft-mozleywilliams-dnsop-dnsaid/.
8+
39
## Supported Versions
410

511
| Version | Supported |

docs/api-reference.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -295,7 +295,7 @@ for agent in result.agents:
295295

296296
### verify()
297297

298-
Verify DNS-AID records for an agent with security validation.
298+
Verify DNS-AID records as interpreted by this implementation, with security validation.
299299

300300
```python
301301
async def verify(fqdn: str) -> VerifyResult

docs/architecture.md

Lines changed: 7 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -2,7 +2,7 @@
22

33
## Overview
44

5-
DNS-AID implements the IETF draft-mozleywilliams-dnsop-dnsaid-01 protocol for
5+
This implementation follows the IETF draft-mozleywilliams-dnsop-dnsaid protocol for
66
DNS-based agent discovery. This document covers the key architectural decisions.
77

88
## Relationship to IETF
@@ -192,9 +192,14 @@ that round-trip the same inputs through every surface.
192192
- Not found → endpoint_source = "http_index_fallback"
193193
```
194194

195-
### Future Enhancement: HTTP Index Fallback in DNS Mode
195+
## Future Enhancements (Non-Normative)
196+
196197
These are implementation proposals and are not part of the current IETF draft.
198+
197199
Items in this section may inform future versions of the specification but should not be treated as authoritative.
200+
201+
### Future Enhancement: HTTP Index Fallback in DNS Mode
202+
198203
Currently the two discovery modes are independent — pure DNS never consults the
199204
HTTP index and vice versa. Per the DNS-AID draft, the HTTP well-known endpoint
200205
is a complementary discovery mechanism. A future enhancement should add an

docs/demo-guide.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,6 @@
11
# DNS-AID Demo Guide
22

3-
This guide demonstrates use of the DNS-AID reference implementation for agent discovery workflows. Perfect for conference calls, IETF presentations, and Linux Foundation demos.
3+
This guide demonstrates use of the DNS-AID reference implementation for agent discovery workflows. The underlying specification is developed at the IETF (https://datatracker.ietf.org/doc/draft-mozleywilliams-dnsop-dnsaid/).
44

55
> **Version 0.6.0** - DNSSEC enforcement, DANE full certificate matching, Sigstore release signing, Route53/Cloudflare SVCB param demotion, JWS signatures for application-layer verification, Tier 1 Execution Telemetry SDK, and community rankings.
66

docs/positioning.md

Lines changed: 77 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,77 @@
1+
# DNS-AID: Background and Comparison
2+
3+
> **Non-normative.** This document provides background context comparing the DNS-AID approach to other agent-discovery proposals. It describes positioning of the DNS-AID *specification* (developed at the IETF), not features specific to this reference implementation. The authoritative specification is the IETF draft: https://datatracker.ietf.org/doc/draft-mozleywilliams-dnsop-dnsaid/.
4+
5+
## vs Competing Proposals
6+
7+
| Approach | Problem | DNS-AID Advantage |
8+
|----------|---------|-------------------|
9+
| **ANS (GoDaddy)** | Centralized registry, KYC required, single gatekeeper | Federated — you control your domain, publish instantly |
10+
| **Google (A2A + UCP)** | Discovery via Gemini/Search, payments via UCP | Neutral discovery — no platform lock-in or transaction fees |
11+
| **.agent gTLD** | Requires ICANN approval, ongoing domain fees | Works NOW with domains you already own |
12+
| **AgentDNS (China Telecom)** | Requires 6G infrastructure, carrier control | Works NOW on existing DNS infrastructure |
13+
| **NANDA (MIT)** | New P2P overlay network, new ops paradigm | Uses infrastructure your DNS team already operates |
14+
| **Web3 (ERC-8004)** | Gas fees, crypto wallets, enterprise-hostile | Free DNS queries, no blockchain complexity |
15+
| **ai.txt / llms.txt** | No integrity verification, free-form JSON | DNSSEC cryptographic verification, structured SVCB |
16+
17+
## Feature Comparison
18+
19+
| Feature | DNS-AID | Central Registry | ai.txt |
20+
|---------|---------|------------------|--------|
21+
| **Decentralized** ||||
22+
| **Secure (DNSSEC)** || Varies ||
23+
| **Sovereign** ||||
24+
| **Standards-based** | ✅ (IETF) |||
25+
| **Works with existing infra** ||||
26+
27+
## The Sovereignty Question
28+
29+
> **Who controls agent discovery?**
30+
> - ANS: GoDaddy (US company as gatekeeper)
31+
> - AgentDNS: China Telecom (state-owned carrier)
32+
> - Web3: Ethereum Foundation
33+
> - **DNS-AID: You control your own domain**
34+
>
35+
> DNS-AID preserves sovereignty. Organizations and nations maintain control over their own agent namespaces with no central authority that can block, censor, or surveil agent discovery.
36+
37+
## Google's Agent Ecosystem
38+
39+
Google is building a full-stack agent platform: **A2A** (communication), **UCP** (payments), and **Gemini/Search** (discovery). While A2A is an open protocol, discovery through Google surfaces means:
40+
- Google controls visibility (pay-to-rank)
41+
- Transaction fees via [UCP](https://developers.google.com/merchant/ucp)
42+
- Platform dependency for reach
43+
44+
**DNS-AID complements A2A** by providing neutral, decentralized discovery — find agents anywhere, not just through Google.
45+
46+
## Understanding the .agent Domain Approach
47+
48+
The [Agent Community](https://agentcommunity.org/) is pursuing a `.agent` top-level domain through ICANN's [new gTLD program](https://newgtlds.icann.org/). Here's how the two approaches compare:
49+
50+
**How .agent Domains Would Work:**
51+
1. Apply to ICANN for `.agent` gTLD (~$185,000 application fee)
52+
2. Wait 9-20 months for ICANN approval process
53+
3. Build registry infrastructure (Open Agent Registry, Inc.)
54+
4. Sell `.agent` domains through accredited registrars
55+
5. Users pay annual registration fees (~$15-50/year per domain)
56+
57+
**How DNS-AID Works:**
58+
1. Use your existing domain (you already own `yourcompany.com`)
59+
2. Add DNS-AID records to your zone (`_myagent._mcp._agents.yourcompany.com`)
60+
3. Start discovering and being discovered immediately
61+
62+
| Factor | .agent gTLD | DNS-AID |
63+
|--------|-------------|---------|
64+
| **Cost to publish** | ~$15-50/year domain fee | Free (use existing domain) |
65+
| **Time to start** | Months (gTLD launch + registration) | Minutes |
66+
| **Who controls discovery** | Registry operator | You (your domain) |
67+
| **Works today** | ❌ Pending ICANN approval | ✅ Works now |
68+
| **Requires new infrastructure** | ✅ Registry, registrars | ❌ Uses existing DNS |
69+
| **Memorable names** |`myagent.agent` | `_myagent._mcp._agents.example.com` |
70+
71+
**The Friendly Take:**
72+
73+
Both approaches share the goal of making AI agents discoverable. The `.agent` gTLD creates a dedicated namespace that's easy to remember (`mycompany.agent`), while DNS-AID leverages existing infrastructure so you can start publishing agents today.
74+
75+
DNS-AID doesn't require waiting for ICANN approval or paying for new domains—it works with the DNS infrastructure your organization already operates. If you own `example.com`, you can publish agents to `_myagent._mcp._agents.example.com` right now.
76+
77+
*Fun fact: When `.agent` domains become available, DNS-AID records will work on them too! The approaches are complementary.*

0 commit comments

Comments
 (0)