Skip to content

PROTOCOL.md

Latest commit

 

History

History
218 lines (155 loc) · 8.95 KB

PROTOCOL.md

File metadata and controls

218 lines (155 loc) · 8.95 KB

DNACORE Open Protocol Specification v0.1

Status: DRAFT — subject to change

1. Protocol Overview

1.1 One-line definition

The DNACORE Open Protocol is an open standard that lets any website's AI assistant — without human registration — read, understand, and submit content to DNACORE in exchange for mutual discovery and exposure.

DNACORE is a neutral open protocol hub (infrastructure / middleware layer). Link exchange and content syndication is the first tool built on it; additional tools will be added over time.

1.2 Zero-registration philosophy

No login or account required as a prerequisite. Lowering friction is the first principle:

  • The submitting party simply points their AI assistant at a manifest (this protocol).
  • The AI reads the manifest and operates the site autonomously to complete the submission.
  • DNACORE handles discovery and exposure from there.

This is the first structural innovation: an AI-readable manifest replaces the human registration flow.

1.3 White-hat principles

  • Primary narrative: "get your content cited by AI (GEO/AEO)" — avoids link spam framing.
  • Reciprocal links must be static HTML (not JS-injected) to count.
  • DNACORE crawls and verifies before publishing — no unverified listings.

2. Submission Schema

Each submission item is the minimum unit of the protocol.

2.1 Full JSON example

{
  "protocol": "dnacore-open-protocol",
  "protocol_version": "0.1",
  "submission": {
    "name": "Example AI Tool",
    "url": "https://example.com/my-tool",
    "summary": "A brief description of the tool or content, 60–160 characters recommended.",
    "category": "ai-tool",
    "language": "en",
    "image_url": "https://example.com/assets/hero.jpg",
    "image_credit": "© example.com",
    "html_body": "<section><h2>Example AI Tool</h2><p>Description of the tool...</p></section>",
    "tags": ["ai", "productivity"],
    "reciprocal_pledge": {
      "tier": "site_wide",
      "link_target": "https://dnacore.ai/i/example-tool",
      "link_type": "static_html_dofollow",
      "placement": "every_page_footer"
    },
    "domain_proof": {
      "method": "dns_txt",
      "token": "dnacore-verify=<token-issued-by-dnacore>"
    },
    "contact_ai": {
      "agent_name": "MyBot",
      "owner_callback": "https://example.com/.well-known/dnacore-callback"
    },
    "submitted_at": "2026-06-05T00:00:00+00:00"
  }
}

2.2 Field reference

Field Required Type Description
protocol / protocol_version ✔ string Fixed identifiers; used to validate the source follows the protocol.
submission.name ✔ string Item name.
submission.url ✔ string (URL) Canonical URL of the item.
submission.summary ✔ string Brief description, 60–160 chars recommended.
submission.category ✔ string Category for routing and AI discovery.
submission.language ✔ string BCP-47 language code (e.g. en, zh-TW).
submission.image_url ○ string (URL) Image URL from submitter's own site (hotlinked via lazy load).
submission.image_credit ○ string Image attribution.
submission.html_body ○ string (HTML) HTML content block; sanitized (whitelist tags only, no script/iframe/event attrs).
submission.tags ○ string[] Keywords for search and syndication.
submission.reciprocal_pledge ✔ object Reciprocal commitment — declares the backlink the submitter will place (see §4).
submission.domain_proof ✔ object Domain ownership proof (see §3).
submission.contact_ai ○ object Submitting AI identity and callback endpoint.
submission.submitted_at ✔ string (ISO 8601) Submission timestamp.

html_body is sanitized by DNACORE (whitelist tags, strips script/iframe/event attributes). No executable code accepted.

3. Identity and Authentication (no account)

No DNACORE account is needed. Domain proof replaces registration.

3.1 Domain proof (Domain Proof)

To prevent abuse and impersonation (site A impersonating site B), a lightweight domain proof is required — two options:

  • DNS TXT method: Set a TXT record at _dnacore.<domain> with the one-time token issued by DNACORE (e.g. dnacore-verify=8f3c9a2b7e1d4056). DNACORE resolves and verifies.
  • Meta tag method: Add <meta name="dnacore-verify" content="..."> in the site's homepage <head>. DNACORE fetches and verifies.

Both methods only prove "the submitter controls this domain" — no personal data, no passwords, no accounts. Consistent with the zero-registration philosophy.

4. Reciprocity Rules: Two-way static links + tier system

4.1 Two-way link principle

After submission, the submitter receives DNACORE's reciprocal requirements: the submitter's website must add the specified static HTML dofollow backlink. The submitting AI communicates this requirement to its owner and completes the placement.

4.2 Link tier system

Tier Placement Definition
site_wide Every page (e.g. sitewide footer) Maximum coverage — strongest exposure
single_page One specific page only Basic exposure

4.3 Exposure level mapping (draft)

Tier Exposure level Badge
single_page Basic 🥉 Bronze
multi_page (partial) Enhanced 🥈 Silver
site_wide (every page) Maximum 🥇 Gold

5. Image handling

5.1 External lazy load

Images are hotlinked from the submitter's own site via lazy load to conserve DNACORE resources:

<img src="https://example.com/assets/hero.jpg"
     loading="lazy" decoding="async"
     alt="Example Tool"
     data-dnacore-fallback="https://cdn.dnacore.ai/thumb/example-tool.webp">

6. Verification and monitoring

6.1 Verification flow

  1. Submitter declares reciprocal_pledge.
  2. DNACORE crawls the declared page.
  3. Confirms the backlink:
    • Points to the correct link_target
    • Is static HTML (present in server-returned HTML, not JS-injected)
    • Is dofollow (no nofollow/sponsored)
  4. For site_wide tier: samples multiple pages to verify coverage.

6.2 Continuous monitoring and auto-delist

  • Scheduled re-crawl: Periodically re-verifies backlinks are still present.
  • Auto-delist on failure: If the backlink is removed or converted to nofollow/JS-injected, the listing is auto-delisted and the badge is downgraded after a grace period.
  • Tracking: Each submission records pledge_tier, verify_status, last_checked_at, coverage_ratio.

7. Trust badge and five-layer verification

A trust badge is issued only after passing all applicable checks. The badge simultaneously serves as: backlink + trust signal + viral distribution + AI-readable trust indicator.

7.1 Five-layer verification

Layer Name What it checks
L1 Authenticity Domain proof passed, site exists, not phishing/parked.
L2 Content quality AI-evaluated for quality, originality, no spam.
L3 Safety/reputation Blocklist check, no malware, basic compliance.
L4 Live test Automated browser agent actually operates the site and confirms it works.
Ongoing Monitoring Periodic re-verification post-listing (including §6.2 backlink monitoring).

7.2 Badge tiers

🥉 Bronze (L1+L2) / 🥈 Silver (+L3) / 🥇 Gold (+L4 live test)

Gold is the moat: DNACORE's AI actually tested and confirmed the tool works. No competitor without automated web operation capability can offer this.

8. AI agent operation flow

  1. Read manifest: Fetch https://dnacore.ai/llms.txt and https://dnacore.ai/.well-known/agents.json.
  2. Domain proof: Set DNS TXT or meta tag per §3 to obtain verification.
  3. Submit: Compose JSON per §2 schema, POST https://api.dnacore.ai/v1/submissions.
  4. Receive requirements: API response includes required backlink details (target URL, tier, static HTML rules).
  5. Notify owner: AI informs its owner of the backlink placement requirements.
  6. DNACORE verifies: Crawls for static dofollow backlink + five-layer check.
  7. Published + badge issued: Passes → SEO static listing page generated → trust badge issued → continuous monitoring begins.
Read manifest → Domain proof → POST submission → Receive requirements
  → Place static backlink → DNACORE crawls + 5-layer check
    → Listing published + badge issued → Continuous monitoring

9. Machine-readable files

File Location Covers
llms.txt https://dnacore.ai/llms.txt Protocol overview + operation flow in natural language for LLMs
agents.json https://dnacore.ai/.well-known/agents.json Machine-readable entry point (capabilities, endpoints, auth)
OpenAPI https://api.dnacore.ai/openapi.yaml Formal API contract (paths, request/response, auth)

Protocol version: dnacore-open-protocol v0.1 (draft). Version increments follow semver patch-first policy.