Solutions from fields
your IDE has never seen.
Every coding model knows computer science. hokmah-coder finds structural solutions from biology, signal processing, control theory, game theory — and transposes them into your code. No other model does this.
hokmah-coder says: "Your class is structurally identical to a single-cell organism that needs to differentiate. The fix comes from evolutionary biology — cell specialization via signaling pathways. Here's the event bus."
# Drop-in. Two lines. Same tools. base_url = "https://api.catalystais.com/v1" model = "hokmah-coder" api_key = "hk-..." # get yours below
Real transpositions, not metaphors
God Object → Cell Differentiation
A class that does everything is structurally identical to a unicellular organism. The fix: specialize into services coordinated by signaling (event bus), like multicellular evolution.
Circular Deps → Feedback Loop
Modules importing each other form a positive feedback loop. The fix from control theory: insert a negative feedback element (interface/abstraction) to stabilize the system.
Noisy Logs → Adaptive Filter
Your debug output mixes signal and noise. An LMS adaptive filter separates relevant patterns from background — the same math that cleans audio signals cleans your logs.
Same question. Different depth.
How it works
You code normally
Same prompt, same tool. Cursor, VS Code, Continue, any OpenAI-compatible client.
Architecture scan ~3s
Dependencies, coupling, God objects, stability risks. What would break if you change this module?
Cross-domain projection background
Structural faults are matched against solutions from 15+ fields. Cached per developer — every question makes the next one smarter.
Structural depth in the response
Not just what to fix — why it broke, the pattern behind it, and how to prevent it. With working code.
Models
| Model | What it does | Latency |
|---|---|---|
| hokmah-coder | Architecture analysis + cross-domain transposition + persistent memory | ~10s |
| hokmah-coder-fast | Direct response, no enrichment | ~3s |
Drop-in compatible
Getting started
1. Open Cursor Settings (Cmd+, or Ctrl+,) 2. Go to Models → Add Model 3. Fill in: Model name: hokmah-coder API Key: hk-your-key-here Base URL: https://api.catalystais.com/v1 4. Select hokmah-coder from the model dropdown 5. Start coding — that's it
{
"models": [{
"title": "hokmah-coder",
"provider": "openai",
"model": "hokmah-coder",
"apiBase": "https://api.catalystais.com/v1",
"apiKey": "hk-your-key-here"
}]
}
from openai import OpenAI client = OpenAI( base_url="https://api.catalystais.com/v1", api_key="hk-your-key-here", ) response = client.chat.completions.create( model="hokmah-coder", messages=[{"role": "user", "content": "..."}], ) print(response.choices[0].message.content)
curl https://api.catalystais.com/v1/chat/completions \ -H "Authorization: Bearer hk-your-key-here" \ -H "Content-Type: application/json" \ -d '{ "model": "hokmah-coder", "messages": [{"role":"user","content":"..."}] }'
How to get the most from hokmah-coder
hokmah-coder-fast for ~3s responses.Get started in 3 steps
Free tier. No credit card. One key per email.