aid: nefesh-ai
name: Nefesh AI
description: >-
  Nefesh AI (nefesh.ai) sells a B2B "human state" middleware API for AI agents: POST live biometric
  signals from any device — heart rate and HRV from a chest strap, webcam rPPG, voice tone, facial
  expression, text sentiment, plus accepted-but-not-yet-fused glucose, EEG, electrodermal, respiratory,
  movement and sleep fields — to one endpoint (api.nefesh.ai/v1/ingest) and get back a unified stress
  score (0-100) mapped to five states with a machine-readable suggested_action and a closed-loop
  adaptation_effectiveness signal. The same capability is exposed four ways: a REST API with webhooks
  and a device registry, a hosted Streamable-HTTP MCP server (6 tools, two of them keyless
  self-provisioning) listed on the official MCP Registry, an A2A agent (4 skills) discovered from a
  card on mcp.nefesh.ai, and a "Cognitive Compute Router" LLM gateway that rewrites system prompt and
  temperature from the live state for OpenAI- and Anthropic-shaped clients. It also maintains the open
  Human State Protocol (HSP 1.0) JSON Schema and an official npm CLI. Positioned as a General Wellness
  product, not a medical device. No OpenAPI is published for any surface.
image: https://nefesh.ai/icon.svg
url: https://raw.githubusercontent.com/api-evangelist/nefesh-ai/refs/heads/main/apis.yml
x-type: company
x-source: harvest:a2a-registry
x-tier: profiled
x-tier-reason: enrichment pipeline local-v3, 2026-09-19
x-aliases:
- Nefesh
- nefesh.ai
- Nefesh AI Ltd
- Nefesh Human State Agent
position: Producing
specificationVersion: '0.20'
created: '2026-09-19'
modified: '2026-09-19'
tags:
- Company
- Biometrics
- Human State
- Stress Detection
- Affective Computing
- AI Agents
- MCP
- A2A
- LLM Gateway
- Wearables
- Digital Health
- Middleware
apis:
- name: Nefesh Human State API
  description: >-
    REST API on api.nefesh.ai (FastAPI): POST /v1/ingest takes a session_id, an ISO 8601 timestamp and at
    least one of 35+ signal fields and returns the fused state, stress_score, confidence, suggested_action and
    (from the second call) adaptation_effectiveness; GET /v1/state, GET /v1/history (1-60 minute lookback),
    GET /v1/triggers (per-subject trigger memory), POST/GET/DELETE /v1/devices (device registry),
    DELETE /v1/subjects/{subject_id} (GDPR cascading deletion), POST /webhooks/register and /webhooks/retries.
    Device-optimised ingest routes at /ingest/rppg, /ingest/polar, /ingest/garmin. Authenticated with the
    X-Nefesh-Key header (keys prefixed nfsh_). Response format is the open Human State Protocol. No OpenAPI is
    published; api.nefesh.ai/openapi.json and /docs are 404 and a custom HTML page respectively.
  image: https://nefesh.ai/icon.svg
  humanURL: https://nefesh.ai/docs/api
  baseURL: https://api.nefesh.ai
  tags:
  - Biometrics
  - Stress Detection
  - Human State
  - Webhook
  - Device Registry
  tags_raw:
  - Biometrics
  - Stress Detection
  - Human State
  - Webhooks
  - Device Registry
  properties:
  - type: APIReference
    url: https://nefesh.ai/docs/api
  - type: Documentation
    url: https://nefesh.ai/docs
  - type: GettingStarted
    url: https://nefesh.ai/docs/quickstart
  - type: Documentation
    name: POST /v1/ingest full field reference
    url: https://nefesh.ai/docs/v1-ingest
  - type: Documentation
    name: State mapping (score bands and suggested_action)
    url: https://nefesh.ai/docs/states
  - type: JSONSchema
    name: Human State Protocol 1.0 response schema (first-party, Apache 2.0)
    url: json-schema/nefesh-ai-hsp-1.0.schema.json
  - type: Webhooks
    url: asyncapi/nefesh-ai-webhooks.yml
  - type: Webhooks
    name: Webhooks documentation (state_change)
    url: https://nefesh.ai/docs/webhooks
  - type: Authentication
    url: authentication/nefesh-ai-authentication.yml
  - type: ErrorCatalog
    url: errors/nefesh-ai-problem-types.yml
  - type: DataModel
    url: data-model/nefesh-ai-data-model.yml
  - type: SourceCode
    name: Integration examples (curl, Python, JavaScript)
    url: https://github.com/nefesh-ai/nefesh-examples
  - type: X-Health
    url: https://api.nefesh.ai/health
- name: Nefesh Cognitive Compute Router
  description: >-
    LLM gateway on gateway.nefesh.ai that injects the caller's live human state into the request and adjusts
    temperature before forwarding to an upstream model. Three modes: OpenAI-compatible POST /v1/chat/completions,
    Anthropic passthrough POST /v1/messages, and Unified Anthropic (Anthropic format to any OpenAI-shaped backend).
    Eleven documented backends selected with the X-LLM-Backend header; the caller supplies X-Nefesh-Key,
    X-LLM-Key and X-Nefesh-Subject (or legacy X-Nefesh-Session). Responses carry X-Nefesh-Adapted, X-Nefesh-State,
    X-Nefesh-Backend and X-Nefesh-Reason. Degrades gracefully to a plain passthrough when no signal exists. Source
    is published; no OpenAPI (gateway.nefesh.ai/openapi.json is 404).
  image: https://nefesh.ai/icon.svg
  humanURL: https://nefesh.ai/docs/gateway
  baseURL: https://gateway.nefesh.ai
  tags:
  - LLM Gateway
  - Proxy
  - Human State
  - OpenAI-Compatible
  tags_raw:
  - LLM Gateway
  - Proxy
  - Human State
  - OpenAI Compatible
  properties:
  - type: Documentation
    url: https://nefesh.ai/docs/gateway
  - type: SourceCode
    url: https://github.com/nefesh-ai/nefesh-gateway
  - type: X-Health
    url: https://gateway.nefesh.ai/health
- name: Nefesh MCP Server
  description: >-
    Hosted remote MCP server at https://mcp.nefesh.ai/mcp (Streamable HTTP, protocol 2025-06-18, session-id
    based; serverInfo nefesh 1.26.0). Anonymous initialize and tools/list succeed and return six tools with full
    inputSchemas — get_human_state, ingest, get_trigger_memory, get_session_history, request_api_key and
    check_api_key_status. The last two work with no key and let an agent self-provision a free key by email
    verification; the other four call the REST API with the X-Nefesh-Key header and return an in-band error
    without one. Published to the official MCP Registry as ai.nefesh/human-state (server.json in the source
    repo), and listed on Smithery and Glama. No stdio package is shipped.
  image: https://nefesh.ai/icon.svg
  humanURL: https://nefesh.ai/docs/mcp
  baseURL: https://mcp.nefesh.ai/mcp
  tags:
  - MCP
  - AI Agents
  - Human State
  properties:
  - type: MCPServer
    url: mcp/nefesh-ai-mcp.yml
  - type: MCPServer
    name: Live Streamable HTTP endpoint
    url: https://mcp.nefesh.ai/mcp
  - type: X-MCPToolsList
    name: Verbatim tools/list response (6 tools with inputSchema)
    url: mcp/nefesh-ai-mcp-tools.json
  - type: X-MCPServerCard
    name: Served /.well-known/mcp.json self-description (verbatim)
    url: well-known/nefesh-ai-mcp-server-card.json
  - type: Documentation
    url: https://nefesh.ai/docs/mcp
  - type: Documentation
    name: MCP tools reference
    url: https://nefesh.ai/docs/mcp/tools
  - type: SourceCode
    url: https://github.com/nefesh-ai/nefesh-mcp-server
  - type: X-MCPRegistry
    name: ai.nefesh/human-state on the official MCP Registry
    url: https://registry.modelcontextprotocol.io/v0/servers?search=nefesh
- name: Nefesh Human State Agent (A2A)
  description: >-
    A2A agent served from the MCP host: card at https://mcp.nefesh.ai/.well-known/agent-card.json, JSON-RPC 2.0
    endpoint at POST https://mcp.nefesh.ai/a2a supporting message/send, message/sendStream, tasks/get and
    tasks/cancel. Four skills — get-human-state, ingest-signals, get-trigger-memory, get-session-history — the
    same capability set as the MCP tools. Authentication is X-Nefesh-Key or Authorization: Bearer; an
    unauthenticated message/send returns a completed Task whose artifact carries the auth error rather than a
    JSON-RPC error. The card is 0.2-era in shape (schemaVersion/agentVersion/authSchemes, no protocolVersion) and
    grades flavored against A2A 1.0.0.
  image: https://nefesh.ai/icon.svg
  humanURL: https://nefesh.ai/docs/a2a
  baseURL: https://mcp.nefesh.ai/a2a
  tags:
  - A2A
  - AI Agents
  - Human State
  properties:
  - type: AgentCard
    url: a2a/nefesh-ai-a2a.yml
  - type: AgentCard
    name: Served agent card
    url: https://mcp.nefesh.ai/.well-known/agent-card.json
  - type: Documentation
    url: https://nefesh.ai/docs/a2a
  - type: SourceCode
    url: https://github.com/nefesh-ai/nefesh-a2a
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com
- FN: APIs.json
  email: info@apis.io
common:
- type: Website
  url: https://nefesh.ai/
- type: DeveloperPortal
  url: https://nefesh.ai/docs
- type: Documentation
  url: https://nefesh.ai/docs
- type: APIReference
  url: https://nefesh.ai/docs/api
- type: GettingStarted
  url: https://nefesh.ai/docs/quickstart
- type: Pricing
  url: https://nefesh.ai/pricing
- type: SignUp
  url: https://nefesh.ai/signup
- type: TermsOfService
  url: https://nefesh.ai/terms
- type: PrivacyPolicy
  url: https://nefesh.ai/docs/privacy
- type: GitHubOrganization
  url: https://github.com/nefesh-ai
- type: Compliance
  name: Regulatory position — General Wellness product; GDPR/BDSG, BIPA and HIPAA statements (no certifications; SOC 2 "in progress")
  url: https://nefesh.ai/docs/regulatory
- type: X-Enterprise
  url: https://nefesh.ai/enterprise
- type: X-Demo
  name: Live browser demo (rPPG + AI chat, no signup)
  url: https://sandbox.nefesh.ai
- type: LLMsTxt
  url: llms/nefesh-ai-llms.txt
- type: LLMsTxt
  name: Provider-hosted llms.txt
  url: https://nefesh.ai/llms.txt
- type: AgentCard
  url: a2a/nefesh-ai-a2a.yml
- type: MCPServer
  url: mcp/nefesh-ai-mcp.yml
- type: WellKnown
  url: well-known/nefesh-ai-well-known.yml
- type: SecurityTxt
  url: well-known/nefesh-ai-security.txt
- type: JSONSchema
  name: Human State Protocol 1.0 (HSP) schema
  url: json-schema/nefesh-ai-hsp-1.0.schema.json
- type: Packages
  url: packages/nefesh-ai-packages.yml
- type: CLI
  url: cli/nefesh-ai-cli.yml
- type: Conformance
  url: conformance/nefesh-ai-conformance.yml
- type: ErrorCatalog
  url: errors/nefesh-ai-problem-types.yml
- type: Lifecycle
  url: lifecycle/nefesh-ai-lifecycle.yml
- type: Authentication
  url: authentication/nefesh-ai-authentication.yml
- type: Conventions
  url: conventions/nefesh-ai-conventions.yml
- type: Sandbox
  url: sandbox/nefesh-ai-sandbox.yml
- type: Plans
  url: plans/nefesh-ai-plans-pricing.yml
- type: RateLimits
  url: rate-limits/nefesh-ai-rate-limits.yml
- type: DataModel
  url: data-model/nefesh-ai-data-model.yml
- type: Webhooks
  url: asyncapi/nefesh-ai-webhooks.yml
- type: X-RegulatoryPosture
  url: regulatory/nefesh-ai-regulatory-posture.yml
- type: DataSubjectRequest
  url: regulatory/nefesh-ai-regulatory-posture.yml
- type: DataResidency
  url: regulatory/nefesh-ai-regulatory-posture.yml
- type: VulnerabilityDisclosure
  url: security/nefesh-ai-vulnerability-disclosure.yml
- type: Security
  name: RFC 9116 security.txt — Contact security@nefesh.ai, Policy https://nefesh.ai/terms
  url: https://nefesh.ai/.well-known/security.txt
- type: DomainSecurity
  url: security/nefesh-ai-domain-security.yml
- type: X-Robots
  url: https://nefesh.ai/robots.txt
- type: X-Sitemap
  url: https://nefesh.ai/sitemap.xml
x-enrichment:
  date: '2026-09-19'
  status: enriched
  artifacts_added: 24
  pass: local-v3