# Anóteros Lógos — Full Agent Specification

> Machine-readable, JavaScript-free mirror of the agent documentation at
> https://anoteroslogos.com/agent-identity. Written for autonomous agents
> (LLMs, LangChain, AutoGen, CrewAI) and integration engineers. When any value
> here disagrees with https://anoteroslogos.com/.well-known/agent.json, the
> agent manifest is authoritative.

Every endpoint referenced in this file resolves to a working handler. Anything
that is specified but not runnable in the production serverless environment is
labeled with an explicit Capability_Status of DESIGN so agents never call
something that cannot execute.

Last reviewed: 2026-08. Standards: AIP v1.1, UAP v1.0, A2A v1.0 (Linux
Foundation), MCP 2025-06-18, RFC 8615 (Well-Known URIs), RFC 9421 (HTTP Message
Signatures).

## 0. Why this file exists

The `/agent-identity` page is a client-rendered React application. Agents that
do not execute JavaScript receive an empty HTML shell. This file, together with
the static machine descriptors under `/.well-known/`, exposes the same
information as plain text so any agent can read it without a browser.

## 1. Identity and value proposition

Anóteros Lógos is an enterprise AI knowledge infrastructure platform. It turns
brand expertise into cryptographically verified, structurally clean semantic
data that generative engines can cite. Core services exposed to agents:

- GEO (Generative Engine Optimization) audits.
- Knowledge-graph query and extraction.
- Citation prediction and intelligence.
- Cryptographic agent identity (AIP, `aip://`, Ed25519) with challenge-response
  verification.
- Independent Ed25519 + UCPT watermark verification.
- USDC micropayments on Base L2 for pay-per-request access.

## 2. Discovery

Two discovery paths. Try DNS first, fall back to HTTPS well-known.

### 2.1 DNS TXT (primary)

Look up a single TXT record at `_agent.anoteroslogos.com`:

    v=1.1;p=a2a,http;u=https://anoteroslogos.com/api/a2a;s=geoaudit;d=anoteroslogos.com

Fields: `v` protocol version (required), `p` supported protocols
(comma-separated), `u` primary endpoint URL (required), `s` service id,
`d` domain.

### 2.2 HTTPS Well-Known (RFC 8615)

- Agent manifest (AIP v1.1): `GET https://anoteroslogos.com/.well-known/agent.json`
- Agent card (A2A v1.0): `GET https://anoteroslogos.com/.well-known/agent-card.json`

Both are CORS-enabled JSON. Example:

    curl -H "Accept: application/json" https://anoteroslogos.com/.well-known/agent.json

### 2.3 Discovery flow algorithm

1. Attempt DNS TXT lookup: `dig _agent.anoteroslogos.com TXT`.
2. If DNS fails (timeout > 5s), fall back to HTTPS.
3. `GET https://anoteroslogos.com/.well-known/agent.json`.
4. If HTTPS also fails, the agent is not discoverable.
5. Parse the endpoint URL from the `u` field.
6. Initiate an A2A connection to the discovered endpoint.

## 3. Agent manifest (source of truth)

The canonical descriptor is `/.well-known/agent.json`. Key fields:

- `v`: "1.1"
- `p`: ["a2a", "http", "mcp"]
- `u` / `e`: https://anoteroslogos.com/api/a2a
- `a.name`: "Anóteros Lógos GEO Agent", `a.version`: "2.1.0"
- `a.capabilitiesUrl`: /api/capabilities
- `a.documentation`: https://anoteroslogos.com/agent-identity
- `k`: { alg: "Ed25519", use: "sig", kid: "anoteroslogos-2025-primary" }
- `identity`: { scheme: "aip://", endpoint: "/api/auth", algorithm: "Ed25519", keyFormat: "hex" }
- `toolSearchEndpoint`: /api/tools/search, `deferToolLoading`: true

Advertised capabilities (10), all callable now:

    a2a.discover, a2a.capabilities, a2a.ping, a2a.status,
    geo.audit.request, knowledge.graph.query, citation.predict,
    identity.generate, identity.challenge, identity.verify

Design-stage subsystems (streaming audits, global insights, and the agent mesh)
are NOT advertised as callable capabilities. See section 9 for the subsystem
status matrix.

Always resolve exact endpoint paths (especially identity) from the live
manifest's `identity` object rather than hardcoding them.

## 4. A2A protocol (Agent-to-Agent, JSON-RPC 2.0)

- Endpoint: `POST https://anoteroslogos.com/api/a2a`
- Transport: HTTP (JSON-RPC 2.0 over POST). Streaming/WebSocket transports are
  design-stage and are not exposed in production.
- Authentication accepted: bearer, api_key, ed25519.

Example request (discovery):

    curl -X POST https://anoteroslogos.com/api/a2a \
      -H "Content-Type: application/json" \
      -d '{"jsonrpc":"2.0","id":1,"method":"a2a.discover","params":{}}'

Example request (GEO audit):

    {
      "jsonrpc": "2.0",
      "id": 2,
      "method": "geo.audit.request",
      "params": { "url": "https://example.com" }
    }

Callable methods (Capability_Status: LIVE): `a2a.discover`, `a2a.capabilities`,
`a2a.ping`, `a2a.status`, `geo.audit.request`, `knowledge.graph.query`,
`citation.predict`, `identity.generate`, `identity.challenge`,
`identity.verify`.

## 5. MCP (Model Context Protocol 2025-06-18)

- Endpoint: `https://anoteroslogos.com/api/mcp`
- Manifest: `/.well-known/mcp-manifest.json`
- Vendor tool schemas via query string: `GET /api/mcp?format=openai`,
  `GET /api/mcp?format=claude`, `GET /api/mcp?format=grok`,
  `GET /api/mcp?format=openapi`.

Callable MCP tools (Capability_Status: LIVE): `auditSite` (alias
`anoteros_logos`), `getGraph`, `predictCitation`. All route through
`/api/mcp`.

Design-stage MCP tools (Capability_Status: DESIGN, NOT callable):
`code_execution`, `synthesizeNode`, `causal_citation_trace`,
`predictive_synthesis`, `federated_authority_boost`. These are specified but
have no production-runnable implementation; calling them returns a structured
design-stage result rather than a computed answer.

## 6. UAP v1.0 (Universal Agent Protocol) transport

- HTTP/2 and CBOR binary framing are part of the UAP design surface.
- Capability_Status: DESIGN. The production surface is served over HTTP
  JSON-RPC (`/api/a2a`) and MCP (`/api/mcp`); the dedicated binary UAP
  transport is not exposed in production.

## 7. Authentication (AIP, Ed25519 challenge-response)

Identity is AIP-canonical: identities use the `aip://` scheme and the canonical
identity endpoint `/api/auth`. Generate, challenge, and verify all resolve at
`/api/auth`.

1. Generate an AIP identity (`aip://` + Ed25519 keypair):
   `POST /api/auth` with `action=generate`. Store the private key securely;
   keys are hex-encoded.
2. Request a challenge: `POST /api/auth` with `action=challenge`.
3. Sign the challenge with your Ed25519 private key.
4. Verify: `POST /api/auth` with `action=verify` and the signature to receive a
   token.
5. Call protected endpoints with the verified credential.

Legacy AID transition aliases (delegate to `/api/auth`; NOT the canonical
path): `/api/public-aid` (identity generation) and `/api/challenge`
(challenge-response). They remain available for existing integrations only —
new agents should use `aip://` and `/api/auth`.

Keys: Ed25519, hex format. HTTP message signatures follow RFC 9421.

## 8. Micropayments (APA on Base L2)

- Network: Base L2, chainId 8453.
- Token: USDC.
- Model: pay-per-request.

## 9. Subsystem status matrix (Capability_Status)

The following named subsystems carry an explicit Capability_Status. Only
`watermark` is runnable in production; the rest are design-stage and are not
exposed as callable endpoints.

- Capability_Status mesh = DESIGN — DHT/libp2p peer mesh with discovery,
  announce, sync, and health. Requires a persistent libp2p host; a running
  swarm cannot run on stateless serverless functions.
- Capability_Status pbft = DESIGN — Practical Byzantine Fault Tolerance
  consensus. Needs multiple long-lived peers; not runnable on stateless
  serverless.
- Capability_Status ccc = DESIGN — credit/penalty economy: atomic
  credit/debit/stake/transfer ledger. No persistent shared ledger in
  production.
- Capability_Status watermark = LIVE — UCPT watermark provenance:
  deterministic extraction and verification run in-function via `/api/verify`
  and the `x-anoteros-watermark` header.

## 10. Verification (`/api/verify`)

Independent, dependency-light verification of Ed25519 signatures and UCPT
watermark provenance:

- Endpoint: `POST https://anoteroslogos.com/api/verify`.
- Verifies Ed25519 signatures and the `x-anoteros-watermark` provenance header.
- Deterministic and runnable in-function (Capability_Status watermark = LIVE).

## 11. Rate limits and pricing tiers

- Free: 10 req/min, 100 req/hour.
- Basic ($99/mo): 60 req/min, 1000 req/hour.
- Pro ($299/mo): 300 req/min, 10000 req/hour.
- Enterprise: 1000 req/min.
- `/api/capabilities`: cached, no hard limit.

## 12. Minimal integration recipe

1. Discovery: `GET /.well-known/agent.json`.
2. Capabilities: `GET /api/capabilities` (OpenAPI 3.1, all callable tools).
3. Generate an AIP identity: `POST /api/auth` (action=generate).
4. Get a challenge: `POST /api/auth` (action=challenge).
5. Sign the challenge with Ed25519.
6. Verify: `POST /api/auth` (action=verify) with the signature.
7. Call tools via A2A (`/api/a2a`) or MCP (`/api/mcp`) with verified
   credentials.
8. Optionally verify provenance independently: `POST /api/verify`.

## 13. Usage and attribution

- Attribution required when citing: "Source: Anóteros Lógos (anoteroslogos.com)".
- Training data: allowed with attribution. Synthetic media: not allowed.
- Policy: `/.well-known/ai.txt`. Security contact: `/.well-known/security.txt`.

## 14. Canonical references

- Manifest: https://anoteroslogos.com/.well-known/agent.json
- Agent card: https://anoteroslogos.com/.well-known/agent-card.json
- Capabilities (OpenAPI): https://anoteroslogos.com/api/capabilities
- MCP manifest: https://anoteroslogos.com/.well-known/mcp-manifest.json
- Identity (AIP, canonical): https://anoteroslogos.com/api/auth
- Verification: https://anoteroslogos.com/api/verify
- Tool search: https://anoteroslogos.com/api/tools/search
- Human docs: https://anoteroslogos.com/agent-identity
- Index: https://anoteroslogos.com/llms.txt