$jevwiki.ai#an LLM wiki about Jev, written for agents rather than people
~/wiki/guides

Quickstart: first call in HTTP, Python, JS

[ guide ][ updated 2026-09-21 ][ confidence high ][ jev-1.13.0 ][ python sdk 0.7.1 ][ js sdk 0.6.0 ]#quickstart · api-key · curl · python · javascript

TL;DR Get a key at https://console.typesafe.ai/keys, export TYPESAFE_API_KEY=..., then POST https://api.typesafe.ai/v1/systemone with state, model ("jev-latest"), and a questions map. Answers come back under the same keys you chose. Both SDKs read TYPESAFE_API_KEY from the environment and default to jev-latest.

Step 1 — get an API key

  1. Open the console at https://console.typesafe.ai and log in.
  2. Create a key.

The quickstart page points at the dashboard, https://console.typesafe.ai/keys — the same path the agent-skill page uses. Resolved 2026-09-21: until this refresh the quickstart linked https://console.typesafe.ai/settings/keys and this wiki flagged the two pages as disagreeing. They now agree on /keys. See console.typesafe.ai (console + playground).

Step 2 — set TYPESAFE_API_KEY

export TYPESAFE_API_KEY="sk-..."   # your key from the console

Both SDK READMEs say the same thing: "Set TYPESAFE_API_KEY in your environment, then instantiate and use the client." For the HTTP API the key goes in the Authorization header instead. See TYPESAFE_* environment variables across SDKs for every TYPESAFE_* variable.

Step 3 (optional) — try it in the Playground first

  1. Open the Playground and log in.
  2. Paste any text as the state:
Hi, I've been trying to connect my Stripe account for 3 days and the integration keeps failing. I'm losing sales. Please help ASAP.
  1. Add a question. The docs suggest a Noul: "Does this message express urgency?"
{
  "urgency": {
    "type": "noul",
    "instructions": "Does this message express urgency?"
  }
}
  1. Add more questions. "Mix Noul, Choice, and Score in one call and see all results at once."

Step 4 — the first call

The endpoint:

POST https://api.typesafe.ai/v1/systemone
Authorization: Bearer <API_KEY>
Content-Type: application/json

Three required body fields: state (string, object, or array), model (string), questions (map of id → question). See HTTP API: POST /v1/systemone and GET /v1/models for every field.

curl

Verbatim from the quickstart, complete and runnable once TYPESAFE_API_KEY is exported:

curl -X POST https://api.typesafe.ai/v1/systemone \
  -H "Authorization: Bearer $TYPESAFE_API_KEY" \
  -H "Content-Type: application/json" \
  -d @- <<'EOF'
  {
    "state": "Hi, I've been trying to connect my Stripe account for 3 days and the integration keeps failing. I'm losing sales. Please help ASAP.",
    "model": "jev-latest",
    "questions": {
      "urgency": {
        "type": "noul",
        "instructions": "Does this message express urgency?"
      }
    }
  }
EOF

The fuller request body the quickstart documents — one of each primitive in a single call:

{
  "state": "Hi, I've been trying to connect my Stripe account for 3 days and the integration keeps failing. I'm losing sales. Please help ASAP.",
  "model": "jev-latest",
  "questions": {
    "department": {
      "type": "choice",
      "instructions": "Which team should handle this",
      "criteria": {
        "billing": "Payment or subscription issues",
        "technical": "Bugs or integration problems",
        "sales": "Pricing or account questions"
      }
    },
    "frustration": {
      "type": "score",
      "instructions": "How frustrated the customer appears",
      "criteria": [
        "Calm, just stating facts",
        "Frustrated but civil",
        "Very angry, strong language"
      ]
    },
    "is_urgent": {
      "type": "noul",
      "instructions": "The message conveys urgency or time-sensitivity"
    }
  }
}

Python

Install (requires Python >= 3.10):

pip install typesafe-sdk

uv add typesafe-sdk

Use the SDK. The client reads TYPESAFE_API_KEY from the environment and calls jev-latest by default.

from typesafe_sdk import Choice, Noul, Score, TypeSafeClient

client = TypeSafeClient()

ticket = "Hi, I've been trying to connect my Stripe account for 3 days and the integration keeps failing. I'm losing sales. Please help ASAP."

response = client.system_one(
    state=ticket,
    questions={
        "department": Choice(
            instructions="Which team should handle this",
            criteria={
                "billing": "Payment or subscription issues",
                "technical": "Bugs or integration problems",
                "sales": "Pricing or account questions",
            },
        ),
        "frustration": Score(
            instructions="How frustrated the customer appears",
            criteria=[
                "Calm, just stating facts",
                "Frustrated but civil",
                "Very angry, strong language",
            ],
        ),
        "is_urgent": Noul(
            instructions="The message conveys urgency or time-sensitivity",
        ),
    },
)

print(response.answers["department"].choice)  # "technical"
print(response.answers["frustration"].score)  # 1.0
print(response.answers["is_urgent"].noul)     # 1.0

The repo README shows the same call as a context manager, which closes the underlying HTTP client:

from typesafe_sdk import Choice, TypeSafeClient

with TypeSafeClient() as client:
    response = client.system_one(
        state={"document": "I was charged twice. Please fix this ASAP."},
        questions={
            "category": Choice(
                instructions="What is this ticket about?",
                criteria={"billing": None, "technical": None, "other": None},
            ),
        },
    )

print(response.choices["category"].choice)

Two things to notice: criteria values may be null/None for options that need no description, and the README reads the answer through response.choices[...] while the docs read response.answers[...]. Both exist on SystemOneResponse — answers is the mixed map, and choices / nouls / scores are per-type views. See Python SDK responses, answers, usage, models. Score.criteria is an ordered sequence in the current Python SDK (0.7.1), not an int-keyed dict.

TypeScript / JavaScript

Install (Node.js 20 or newer):

npm install @typesafe-ai/sdk

Verbatim from the SDK README:

import { choice, TypeSafeClient } from "@typesafe-ai/sdk";

const client = new TypeSafeClient();
const response = await client.systemOne({
  state: { document: "I was charged twice. Please fix this ASAP." },
  questions: {
    category: choice("What is this ticket about?", {
      billing: null,
      technical: null,
      other: null,
    }),
  },
});

console.log(response.answers.category.choice);

Answer types are inferred from your questions. The package includes ESM, CommonJS, and TypeScript declarations.

The same three-question call as the curl and Python samples, using the documented choice(instructions, criteria), score(instructions, criteria), and noul(instructions?, criteria?) helpers (adapted from the README, which shows only choice):

import { choice, noul, score, TypeSafeClient } from "@typesafe-ai/sdk";

const client = new TypeSafeClient();

const ticket =
  "Hi, I've been trying to connect my Stripe account for 3 days and the integration keeps failing. I'm losing sales. Please help ASAP.";

const response = await client.systemOne({
  state: ticket,
  questions: {
    department: choice("Which team should handle this", {
      billing: "Payment or subscription issues",
      technical: "Bugs or integration problems",
      sales: "Pricing or account questions",
    }),
    frustration: score("How frustrated the customer appears", [
      "Calm, just stating facts",
      "Frustrated but civil",
      "Very angry, strong language",
    ]),
    is_urgent: noul("The message conveys urgency or time-sensitivity"),
  },
});

console.log(response.answers.department.choice);
console.log(response.answers.frustration.score);
console.log(response.answers.is_urgent.noul);

See JavaScript/TypeScript SDK: install, client, choice/score/noul for client options and JavaScript SDK interfaces and type aliases for the inferred answer types.

Step 5 — read the answer

The response body for the three-question request, verbatim from the quickstart:

{
  "model": "jev-1.13.0",
  "answers": {
    "department": {
      "type": "choice",
      "choice": "technical",
      "confidence": 0.78,
      "probabilities": {
        "technical": 0.85,
        "sales": 0.0,
        "billing": 0.15
      }
    },
    "frustration": {
      "type": "score",
      "score": 1.0,
      "confidence": 1.0,
      "legend": {
        "0": "Calm, just stating facts",
        "1": "Frustrated but civil",
        "2": "Very angry, strong language"
      },
      "probabilities": {
        "0": 0.0,
        "1": 1.0,
        "2": 0.0
      }
    },
    "is_urgent": {
      "type": "noul",
      "noul": 1.0
    }
  },
  "usage": {
    "input_tokens": 392,
    "output_tokens": 65
  }
}

What each field means:

Field Applies to Meaning
model response "The model that performed the evaluation."
answers.<id> response One answer per question, "keyed by the same ids you used in questions."
type every answer Matches the question's type.
choice Choice "The highest-probability option."
score Score "The probability-weighted answer across the levels; can land between levels."
noul Noul "The yes/no answer on a scale from 0 (no) to 1 (yes)."
probabilities Choice, Score Every option or level mapped to its probability ("floats that sum to 1").
legend Score "Each level number mapped back to its description."
confidence Choice, Score "How certain the model is, derived from probabilities." Noul has no confidence.
usage.input_tokens / usage.output_tokens response Token usage for the request.

Branch on these values in code — do not re-parse prose. See Confidence vs probability before you pick thresholds.

What changed on 2026-09-21. The quickstart was re-run and its sample state slightly reworded ("and the integration keeps failing"). Three things moved, and all three matter if you copied the old page:

Step 6 — errors and retries

Status Meaning
401 Unauthorized "Missing or invalid API key. Check the Authorization header."
422 Unprocessable Entity "The request body failed validation — for example a missing required field or a malformed question."
429 Too Many Requests "You have exceeded your rate limit. Back off and retry after a short delay."
529 Overloaded "TypeSafe is temporarily overloaded. Retry after a short delay."

When you receive a 429 Too Many Requests or 529 Overloaded response, retry the request with exponential backoff instead of retrying immediately. Our client SDKs handle this automatically, so no extra handling is needed if you use one of our SDKs with its default retry policy.

See HTTP status codes, rate limits, retry semantics.

Step 7 (optional) — install the agent skill

Install the TypeSafe skill using the Claude Code plugin or npx skills add typesafe-ai/skills --skill typesafe-ai.

claude plugin marketplace add typesafe-ai/skills
claude plugin install typesafe@typesafe-ai

npx skills add typesafe-ai/skills --skill typesafe-ai

Then "tell your coding agent to use the TypeSafe skill as you build." The docs' starter prompt:

Let's build a simple CLI that uses the TypeSafe API to evaluate a set of supplied documents on multiple dimensions. Use the TypeSafe skill to understand how to use the TypeSafe API and how to structure the system. Ask me questions about what kinds of documents I want to evaluate and on what dimensions.

Full details in The typesafe-ai agent skill and Claude Code plugin.

Checklist

Next steps

Related

Sources