# TypeSafe AI

**Canonical:** https://apis.io/providers/typesafe-ai/  
**Website:** https://typesafe.ai/  
**APIs profiled:** 4

TypeSafe AI is a San Francisco AI lab building System One models — a class of model trained to return typed, calibrated decisions for software instead of generated text. Its flagship and first System One model, Jev, is served by a single HTTP endpoint: POST /v1/systemone takes a `state` (a string, JSON object, or array) plus a map of typed questions and returns one structured answer per question. Three question primitives are published: Noul (a yes/no question answered with the probability the answer is yes), Choice (one option from a caller-defined set, returned with the full probability distribution) and Score (a rating against ordered rubric levels, returned as a probability-weighted value plus a legend). Choice and Score answers also carry a confidence value derived from the distribution, which is the mechanism TypeSafe intends callers to threshold on to decide when code may act autonomously and when it must escalate to a human. The model is trained with Reinforcement Learning for Calibrated Decisions (RLCD) rather than RLHF, is not fine-tuned or LoRA-adapted per customer, and is not trained on customer requests or responses. The company publishes an OpenAPI 3.1 contract on its API host, official Python and TypeScript SDKs, an installable Agent Skill for Claude Code and other coding agents, an A2A agent card, an anonymous documentation MCP server, an llms.txt index, a Better Stack status page and a Vanta trust center.

## Kin Score — 52.1 / 100 (developing)

Scored 2026-10-09 under rubric 0.23.0. Trend: flat (+0.5 from 51.6).

| Facet | Score |
|---|---|
| Discoverability | 80.8 |
| Contract Quality | 48.3 |
| Contract Governance | 18.2 |
| Operational Transparency | 36.8 |
| Developer Ergonomics | 69.0 |
| Access Clarity | 35.5 |

Regulatory layer — **Horizontal (data, software, accessibility, platform)**: 47.3 (matched via fallback).

## Agent readiness — 37.8 (agent-ready)

| Dimension | Value |
|---|---|
| Spec Presence | yes |
| Agentic Access | no |
| Reversibility Documented | no |
| MCP Server | documented |
| Auth Clarity | bearer |
| Idempotency | no |
| Error Semantics | documented |
| OpenAPI Examples | documented |
| Rate Limit Signal | documented |
| Event Surface Described | no |
| Agent Skills | yes |
| Well Known Catalog | yes |
| Consent Identity | no |
| Agent Card | conformant |
| Dry Run Mode | no |
| Delegated Identity | no |
| Protected Resource Metadata | no |
| Dynamic Client Registration | no |
| Agentic Commerce | no |

## APIs (3)

- **TypeSafe AI Documentation MCP Server** — Hosted, anonymous, read-mostly MCP server over the published TypeSafe documentation corpus. Three tools: search_type_safe_ai (knowledge-base search), query_docs_filesystem_type_...
- **TypeSafe AI Models API** — The Models API from TypeSafe AI — 1 operation(s) for models.
- **TypeSafe AI Systemone API** — The Systemone API from TypeSafe AI — 1 operation(s) for systemone.

## MCP servers (1)

- **TypeSafe AI** — TypeSafe runs a live, anonymous, hosted MCP server at https://docs.typesafe.ai/mcp. It answered a cold JSON-RPC tools/list with no credentials and returned three real tools with...

## Security (4)

- **Typesafe Ai Authentication** — http · 1 scheme
- **Typesafe Ai Domain Security** — TLSv1.3 · HSTS · DNSSEC · DMARC
- **Typesafe Ai Vulnerability Disclosure** — Hackerone · contact published
- **Typesafe Ai Trust Center** — observed, observed_note, how_to_close

## Plans (1)

- **Typesafe Ai Plans Pricing**

## Tags

Artificial Intelligence, Machine Learning, Classification, Content Moderation, Decision Support, Structured Outputs, Inference, LLM Alternative, Agent Skills, MCP, Agent-Native, Developer Tools, A2A

---

Profiled by [API Evangelist](https://apievangelist.com) and published on [APIs.io](https://apis.io/providers/typesafe-ai/). Scores are computed from the provider's own public artifacts under a published rubric.