ISMS Copilot

Last updated: 2026-09-03 · Audience: developers and product engineers calling the ISMS Copilot API directly

OpenAI-compatible compliance API quickstart

Your OpenAI SDK already knows the wire. Change the base URL, use an sk-isms key, and you get compliance-literate completions with curated framework knowledge injected server-side. Compatibility is a documented text-only chat completions subset, not the whole OpenAI API. This page walks the first request end to end.

The short version

  • Base URL: https://api.ismscopilot.com/v1
  • Auth: Authorization: Bearer sk-isms-... (create at platform.ismscopilot.com/keys)
  • Models: isms-fast, isms-thinking, and EU-oriented twins isms-fast-eu, isms-thinking-eu.
  • Shape: documented text-only chat completions subset. No tool calling.
  • Do not invent prices here. Prepaid credits, minimum top-up, and unit rates live on platform.ismscopilot.com/pricing.

Why a dedicated quickstart

The API docs cover every detail; this page is the shortest path from zero to a working call, with the exact commands you can paste. If you are still deciding whether to call the API at all versus embedding a finished widget or building in-house, read the build vs buy decision guide first. If the question is why this API instead of any model, read the vs-any-model argument. If you are wiring a coding agent, use the compliance sub-agent guide. This page assumes you already picked the API and want a first completion.

1. Create an API key

  1. Open platform.ismscopilot.com/keys and create a new key.
  2. Copy it immediately. It starts with sk-isms- and is shown only once. Treat it like a password; revoke and rotate if it leaks.
  3. Optionally set per-key spend caps (hourly, daily, weekly, monthly) so a leaked key or a runaway integration cannot empty your balance.
  4. Make sure the key has a positive prepaid credit balance. Completions fail against a zero balance until you top up.

2. First request in curl

Standard chat completions shape. Swap in your key and question.

curl https://api.ismscopilot.com/v1/chat/completions \
  -H "Authorization: Bearer sk-isms-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "isms-fast",
    "messages": [
      {"role": "user", "content": "List 3 ISO 27001 access control requirements."}
    ]
  }'

Omit max_tokens for the default safety behavior, or set it to at least 1024 if you use a thinking alias and want to cap length yourself.

3. Same request with the OpenAI SDK

Point the official OpenAI SDK at the ISMS Copilot base URL. For the documented text-only chat completions subset, that is the only integration change. Do not send tool definitions.

Python

from openai import OpenAI

client = OpenAI(
    base_url="https://api.ismscopilot.com/v1",
    api_key="sk-isms-...",
)

resp = client.chat.completions.create(
    model="isms-thinking",
    messages=[{"role": "user", "content": "Difference between SOC 2 Type 1 and Type 2?"}],
)
print(resp.choices[0].message.content)

Node.js

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.ismscopilot.com/v1",
  apiKey: "sk-isms-...",
});

const resp = await client.chat.completions.create({
  model: "isms-fast",
  messages: [{ role: "user", content: "Which GDPR articles cover the DPO?" }],
});
console.log(resp.choices[0].message.content);

Model aliases

Four aliases, two intents (fast vs thinking) each on two paths (global vs EU-oriented).

AliasPathIntent
isms-fastGlobalFast answers for the common case: lookups, drafting, quick control questions.
isms-thinkingGlobalDeeper reasoning for multi-step or ambiguous questions. Typically uses more tokens than Fast for the same question; unit rates are on the platform pricing page.
isms-fast-euEU-orientedFast alias routed through the EU inference path.
isms-thinking-euEU-orientedThinking alias routed through the EU inference path.

Full detail: docs.ismscopilot.com/docs/api/models-and-regions.

Reading the live frameworks catalog

Default auto-detection scans user messages and the last assistant turn, not system or developer messages. Pin catalog ids when you already know the framework. To list what the catalog currently covers, call the public endpoint directly:

curl https://api.ismscopilot.com/v1/frameworks

No key required for this endpoint. It returned 113 entries on 2026-09-03, each with an id, name, region, and version. Treat the live response as authoritative; do not hard-code the count.

Data retention and credits, briefly

Zero Data Retention

Request content is not stored as customer records. Only usage and billing metadata (tokens, cost, model, status) is kept, so you can meter prepaid credits. Read the full posture, including upstream provider detail, at docs/api/zero-data-retention before you cite it in a customer contract.

Prepaid credits

Usage debits prepaid credits, separate from any chat subscription. A zero balance returns a rate-limit / quota style error until you top up. Current minimum top-up and unit rates live on platform.ismscopilot.com/pricing; how credits and spend caps work is documented at docs/api/credits-and-limits.

Next steps

  1. 1. Set a spend cap. Do this before you wire the key into anything that calls it in a loop.
  2. 2. Read the system prompt page. If you need to understand exactly what framing the model receives, docs/api/system-prompt publishes it in full.
  3. 3. Wire it as a sub-agent. Tool-using coding agents should call this endpoint as a compliance step, not as their global model. See the compliance sub-agent guide. For why this API instead of any model, see the vs-any-model argument.
  4. 4. Decide API vs Embed. If what you actually want is a finished chat widget rather than raw completions, see ISMS Copilot Embed instead.
  5. 5. Keep accountability clear. A compliance-literate completion is not a certification decision. Pair this with the agent accountability guide if your product surfaces answers to end customers.

Frequently asked questions

Is it really OpenAI-compatible?

Yes, as a documented text-only subset. The chat completions endpoint accepts the same core request shape and returns the same response shape as the OpenAI API, so official and community SDKs work by changing the base URL and the API key. Tool or function calling, JSON mode, n above 1, logprobs, and multimodal parts are rejected. Streaming is supported. Details: docs.ismscopilot.com/docs/api/get-api-key.

Which model alias should I call first?

isms-fast for most product traffic. isms-thinking when the question needs multi-step reasoning across a framework. Add the -eu suffix (isms-fast-eu, isms-thinking-eu) if you need the EU-oriented inference path. Do not hard-code assumptions about which is cheaper; check platform.ismscopilot.com/pricing.

Do I need to tell it which framework I'm asking about?

Default auto-detection scans named frameworks in user messages and the last assistant turn, not system or developer messages. You can also pin catalog ids (max 8) or send none. Every successful completion discloses which modules were injected. Pin when you already know the framework; do not rely on detection to read a system prompt. Details: docs.ismscopilot.com/docs/api/framework-knowledge.

What does it cost?

This page does not state unit rates or credit amounts. The API runs on prepaid credits, separate from any chat subscription, with a minimum top-up and optional per-key spend caps (hour, day, week, month) so a bug cannot empty a wallet. platform.ismscopilot.com/pricing is the source for current numbers; docs.ismscopilot.com/docs/api/credits-and-limits explains how credits and spend caps work.

What happens when my credit balance hits zero?

Requests fail with a rate-limit / quota style error until you top up again. Nothing silently degrades or queues; check platform.ismscopilot.com/usage if a call fails unexpectedly.

Do you store my prompts and completions?

No. Zero Data Retention applies to request content: prompts and model outputs are not stored in the API data layer, and upstream paths (OpenRouter zero-retention endpoints globally, Mistral under a ZDR agreement for the EU aliases) are configured the same way. Usage history in the console shows metadata only (model, tokens, cost, status), retained for metering and billing. Read docs.ismscopilot.com/docs/api/zero-data-retention before you put that claim in a customer contract.

Is this the same as ISMS Copilot for Agents (MCP)?

No. This API gives your code raw compliance-literate completions with sk-isms keys and prepaid credits; you build the product surface. ISMS Copilot for Agents (Account MCP) gives an AI tool like Claude Code or Codex access to your own account (conversations, workspaces, memories) with pat-isms tokens on your chat subscription. Different keys, different jobs.

How many frameworks does the catalog cover?

The public catalog returned 113 entries via GET https://api.ismscopilot.com/v1/frameworks on 2026-09-03. The registry updates over time; call the live endpoint rather than hard-coding a count in your integration or pitch deck.

Primary sources

  • ISMS Copilot API documentation (live product truth). docs.ismscopilot.com (checked 2026-09-03).
  • Get an API key and first completion (steps, curl, Python, Node examples). docs.ismscopilot.com (checked 2026-09-03).
  • Models and regions (alias list, EU vs global paths). docs.ismscopilot.com (checked 2026-09-03).
  • Framework knowledge (how detection works). docs.ismscopilot.com (checked 2026-09-03).
  • Zero Data Retention (request content posture, what is retained as metadata). docs.ismscopilot.com (checked 2026-09-03).
  • Credits, pricing, and spend limits. docs.ismscopilot.com (checked 2026-09-03).
  • Public frameworks catalog (GET /v1/frameworks; count verified on this page's last-updated date). api.ismscopilot.com (checked 2026-09-03).
  • OpenAI API reference for chat completions (industry shape this endpoint targets). platform.openai.com (checked 2026-09-03).

Written and maintained by the ISMS Copilot team for API integrators. Last reviewed 2026-09-03.

This is a developer quickstart, not a pricing quote or legal advice. Product capabilities, retention posture, and commercial terms are defined by live docs, the Trust Center, and the platform console. Where this page summarizes those surfaces, the live surface wins on drift.