AgentFoundry

AgentFoundry is Syntheticore's proprietary Python library for building and orchestrating enterprise AI agents. It provides LLM orchestration, governed tool execution, multi-tenant memory, vector and knowledge-graph integration, workflow lifecycle management, and execution tracing.

Start Here

You want	Read
Coding standards, TDD, IP rules, and agent workflow	`AGENTS.md`
Documentation map and current priorities	`docs/README.md`
Current backlog	`docs/TODO.md`
Configuration and environment variables	`docs/reference/Configuration_Guide.md`
Supported LLM providers	`docs/reference/LLM_Providers.md`
Memory architecture and operations	`docs/reference/Memory_Management_Guide.md`

What It Provides

Flat, hierarchical, and deep-agent orchestration over LangGraph
Registry-based tool discovery and governed execution
Thread, user, organization, and global memory scopes
Milvus and FAISS vector-store providers
DuckDB/SQLite, PostgreSQL, and Neo4j knowledge-graph providers
Dependency-aware workflow execution with pause, resume, cancel, retry, and replan
Request tracing, cost accounting, execution budgets, and audit export
Cython-compiled distribution for performance and IP protection
Fail-fast initialization and operation: required backends do not silently degrade to substitute providers or pseudo-embeddings

Requirements

Python 3.11+
The dependencies in requirements.txt
A valid AgentFoundry license for enforced runtime use

Quick Start

Use the top-level public API for host-facing classes and the documented agentfoundry.config API for configuration:

from agentfoundry import Orchestrator, ToolRegistry
from agentfoundry.config import AgentConfig, set_module_config

config = AgentConfig.from_dict({
    "scope": {
        "org_id": "myorg",
        "user_id": "u1",
        "thread_id": "t1",
    },
    "llm": {
        "provider": "openai",
        "model_name": "gpt-5.4",
        "api_key": "sk-...",
    },
    "vector_store": {
        "provider": "milvus",
        "milvus": {"uri": "http://localhost:19530"},
    },
})
set_module_config(config)

registry = ToolRegistry(config=config)
registry.load_tools_from_directory()
orchestrator = Orchestrator(registry, config=config)

result = orchestrator.process_request(
    "Summarize recent activity",
    config={
        "configurable": {
            "org_id": "myorg",
            "user_id": "u1",
            "thread_id": "t1",
        }
    },
)

For a TOML deployment, start from agentfoundry.template.toml:

config = AgentConfig.from_toml("agentfoundry.toml", apply_env=True)

Canonical environment overrides use nested AF__* paths, for example AF__LLM__PROVIDER and AF__VECTOR_STORE__MILVUS__URI. See the Configuration Guide for precedence, secret aliases, and migration from legacy flat keys.

Development

AgentFoundry follows mandatory test-driven development. Most production modules are Cython compiled, so rebuild extensions after changing a Python source that has a matching shared object:

python setup.py build_ext --inplace
python -m pytest tests/test_<module>.py -v

See AGENTS.md for the complete workflow and docs/workflows/tdd-workflow.md for the TDD guide.

Licensing

This software is proprietary and confidential. Use, reproduction, modification, distribution, or commercial deployment requires explicit written authorization from Syntheticore, Inc. Licensing inquiries: csteel@syntheticore.com.

Metadata

Release files for agentfoundry 1.5.90

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Built distributions (wheels)

Table of built distributions (wheels) for agentfoundry 1.5.90
File	Interpreter	ABI	Platform	Reset
agentfoundry-1.5.90-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl 7.7 MB Sep 18, 2026	CPython 3.12	CPython 3.12	Linux glibc 2.17+ x86-64	Details
agentfoundry-1.5.90-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl 7.6 MB Sep 18, 2026	CPython 3.12	CPython 3.12	Linux glibc 2.17+ ARM64, Linux glibc 2.28+ ARM64	Details

Total release size: 15.3 MB

Release files / agentfoundry-1.5.90-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl

Download URL	agentfoundry-1.5.90-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl
Size	7.7 MB
Tags	CPython 3.12 Linux glibc 2.17+ x86-64
SHA-256 checksum How to use checksums	`b3e24abf0ee2ac86f27636a031122046d6980b8613c5e62d717dbe2c25a002d8`
BLAKE2b-256 checksum How to use checksums	`87b54f1e8857cbe3994709a96be6864d3eb4897113e2d4db61a13d7bce281266`
Upload date	Sep 18, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/7.0.0 CPython/3.12.13`

Release files / agentfoundry-1.5.90-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl

Download URL	agentfoundry-1.5.90-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.manylinux_2_28_aarch64.whl
Size	7.6 MB
Tags	CPython 3.12 Linux glibc 2.17+ ARM64 Linux glibc 2.28+ ARM64
SHA-256 checksum How to use checksums	`13eb834ed86dfade7cd6e7015521a579f05caa7b5f71a594d78090877f348ed2`
BLAKE2b-256 checksum How to use checksums	`7af16eb3ef9d9c56f4bbea64628f74b6ab90c36063d44b575ae8637ec719253b`
Upload date	Sep 18, 2026
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/7.0.0 CPython/3.12.13`