CognOs

Why we Built CognOs

"I intend, therefore i am"

This agent framework uses induced schema to simulate learning and generate agents with predictable, mutable and auditable behavior with the goal of alligning to the user.

Agents map spaces, sense spaces then extract information to make decisions.

Running Docker Image

run docker-compose up --build

Tech stack

API Framework

FastAPI

Short Term Memory and Cache Managment

Redis

Long Term Memory

Supabase

supported LLMs

• Anthropic
• Gemini
• OpenAI
• Ollama
• any llm that generates parsable json

API

Agents

Create Your First Agent

curl -X POST http://localhost:8000/agent/create   -H "Content-Type: application/json"   -d '{
        "agent_name": "test",
        "agent_type": "gemini",
        "agent_prompt": "test prompt",
        "schema": {}
      }'

Example: Create an Exrtractor Agent

curl -X POST http://localhost:8000/agent/create   -H "Content-Type: application/json"   -d '{
        "agent_name": "Exctraction Agent",
        "agent_type": "gemini",
        "agent_prompt": "make your output a json i need a output feild, reason feild, Entity List feild consiting of entity and hypernym.",
        "schema": {}
      }'

List your Agents

curl -X GET http://localhost:8000/agent/

Select An Agent

curl -X GET http://localhost:8000/agent/<your-agent-id>

Upload

List upload

curl -X GET http://localhost:8000/upload/list

Upload File

curl -X POST http://localhost:8000/upload/upload \
  -F "file=@./test.pdf" \
  -F "upload_as=my_custom_name.pdf"

Get File

curl http://localhost:8000/upload/get/mydoc.pdf