INTELISYS
Monitor
OverviewRequestsUsageCosts
Build
API KeysEndpointsDocumentation
Account
Credits & plansSettings
·Loading workspace…
INTELISYS
Docs
engine live
…
Documentation
Quick startConnect an agentChat completionsModelsCredits & billingFAQ
Developer

Documentation.

Everything you need to send your first request through the Intelysis engine.

Quick start

Intelysis is an OpenAI-compatible API. Point any coding agent at it and the engine optimizes context, cache, and cost in between — no workflow rewrite, no new SDK.

1

Create a key. Open API Keys and press Create key. Copy the secret — it's shown once.

2

Set your base URL. Point your client at the Intelysis endpoint:

config
base_url = https://app.intelisys.dev
# auth: Bearer <your Intelysis API key>
3

Send a test request. Everything OpenAI-compatible works out of the box:

curl
curl https://app.intelisys.dev/v1/chat/completions \  -H "Authorization: Bearer ***" \  -H "Content-Type: application/json" \  -d '{
    "model": "deepseek-v4-flash",
    "messages": [
      {"role": "system", "content": "You are a senior TypeScript engineer."},
      {"role": "user", "content": "Fix the OAuth callback bug."}
    ]
  }'
Intelysis is BYOK. Add your own DeepSeek key in Settings; DeepSeek bills you directly for inference. Intelysis charges a flat plan fee.

Connect an agent

Any client that speaks the OpenAI chat-completions protocol can route through Intelysis.

OpenCodeOpen the provider config, set base_url to the Intelysis endpoint, and use an Intelysis API key.
ClineProvider settings → API base URL → Intelysis endpoint, set the API key (any name).
CodexOverride the model's base URL to route requests through Intelysis.
HermesSet the endpoint in your Hermes provider config; Hermes keeps its own model and key settings.
AiderUse the OpenAI-compatible base URL and an Intelysis API key in your .aider.conf.yml.
RoomotePoint the Roomote model provider at Intelysis when you run it with your own model key.
Custom clientsSet base URL + Authorization header. That's it.

Chat completions

The main endpoint. The engine optimizes before the request reaches DeepSeek: it prunes context to what the task needs, reuses cached content, and validates the output.

request
POST https://app.intelisys.dev/v1/chat/completions
Authorization: Bearer ***
{
  "model": "deepseek-v4-flash",
  "messages": [{"role": "user", "content": "…"}],
  "nexus_mode": "balanced"        // optional: balanced | saver | ultra (Balanced | Economy | Maximum)
}

Every response body carries an x-nexus-report object with what the engine changed on that request — tokens removed, savings, per-method attribution:

x-nexus-report
{
  "original_tokens": 14257,
  "compiled_tokens": 10771,
  "token_reduction_pct": 24.45,
  "savings_pct": 34.39,
  "optimizations": [
    {"method": "semantic_compression", "tokensRemoved": 3533}
  ]
}

Models

GET /v1/models returns the model the engine routes to. Use that id as the model in chat completions.

curl
curl https://app.intelisys.dev/v1/models \  -H "Authorization: Bearer ***"

Credits & billing

Intelysis is bring-your-own-key. Add your DeepSeek key in Settings; it is encrypted at rest, used only for your workspace, and never returned to the browser. DeepSeek bills you directly for the inference it runs. The plan covers the optimizer, analytics, and support.

·Each request still records its optimized cost so you can compare it with the provider bill.
·Without a provider key, inference is unavailable until you add one.
·No token markup: Intelysis does not resell credits on the BYOK path.

FAQ

Do I need my own DeepSeek API key?+

Yes. Add it once in Settings; it is encrypted at rest and never returned to the browser. DeepSeek bills you directly for inference.

What happens when my plan ends?+

Requests return a 402 subscription_required response with a link to Pricing. A trial does not auto-charge because card checkout is not live yet.

Which agents are supported?+

Anything OpenAI-compatible: OpenCode, Cline, Codex, Cursor, custom scripts. Point the base URL at Intelysis and set your API key.

How are savings calculated?+

Every request records what the same call would have cost at full context vs. what the engine actually sent — the difference is your savings. You see the per-request line item in Requests.

Can I change the optimization profile?+

Yes — use the profile dropdown in the top bar, or set nexus_mode per request. Balanced is the default; Economy and Maximum trade more context for lower spend.