ioc-knowledge-memory 0.2.1

ioc-knowledge-memory-svc

ioc-knowledge-memory-svc

ioc-knowledge-memory-svc - APIs for knowledge management

Usage Options

This service can be accessed in two ways:

1. HTTP API (for external clients, microservices) - Port 9003
2. Direct Library (for Python in-process, 10x faster) - See knowledge_memory

Prerequisites

• Python 3.8+
• Poetry: curl -sSL https://install.python-poetry.org | python3 -
• Task:
  • macOS: brew install go-task
  • Linux: apt install task, dnf install go-task, or snap install task --classic
  • Cross-platform: npm install -g @go-task/cli
  • Go users: go install github.com/go-task/task/v3/cmd/task@latest
  • Manual install: sh -c "$(curl --location https://taskfile.dev/install.sh)" -- -d -b ~/.local/bin
  • All-in-one setup: ./install.sh (installs Poetry and Task globally, and dependencies)

Quick Start

Deployment Options

task docker-compose-up       # Start complete stack (application + databases)

Alternative Quick Start Methods

Dev setup (installs Poetry and Task globally)

./install.sh
task dev

Manual setup (if you have Poetry/Task already)

poetry install
task dev

API Documentation: http://localhost:9003/docs http://localhost:9003/openapi.json

Development

Using Task

task dev              # Start development server
task test             # Run all tests
task docker-build     # Build Docker image
task docker-run       # Run Docker container

Using Poetry directly

cd src/server
poetry run python main.py

Using Docker

docker-compose up --build

Version Management

This project uses pyproject.toml as the single source of truth for versioning.

How to Bump Version

Simply edit one line in pyproject.toml:

[project]
version = "0.1.5"  # Change this line only!

All other locations automatically sync:

• ✅ setup.py - reads from pyproject.toml
• ✅ src/knowledge_memory/__init__.py - uses importlib.metadata
• ✅ .github/workflows/ci.yaml - dynamically reads and updates for dev builds

No manual sync needed! The version is automatically propagated to:

• Package metadata when building
• Runtime __version__ attribute
• CI/CD dev builds (adds .devYYYYMMDDHHMMSS suffix)

Version Format

We follow PEP 440 versioning:

• Release versions: 0.1.4, 0.2.0, 1.0.0
• Dev builds (CI): 0.1.4.dev20260316120000

Python Library (New!)

For in-process, high-performance access without HTTP overhead:

from knowledge_memory import upsert_knowledge_graph, query_knowledge_graph
from knowledge_memory import onboard_vector_store, upsert_vector_store

# Direct function calls - no HTTP!
response = upsert_knowledge_graph(
    mas_id="agent-1",
    wksp_id="workspace-1",
    concepts=[{"id": "c1", "name": "Python"}]
)

Benefits:

• 🚀 10x faster than HTTP API (no network overhead)
• 🐍 Type-safe Python exceptions
• 🔧 Same features as HTTP API
• ✅ No regression - HTTP API still works

Documentation: See src/knowledge_memory/README.md

Quick test:

python src/knowledge_memory/examples.py     # Run examples
pytest tests/test_knowledge_memory.py -v    # Run tests
python3 validate_knowledge_memory.py        # Validate structure