Skip to main content

SophiaGraph logo

SophiaGraph

Durable memory, provenance, and knowledge-workspace primitives for local AI systems.

GitHub · PyPI · Website · Docs · X

PyPI Python License Status

The current SophiaGraph source is the v0.0.7 alpha release line for durable, typed memory. It stores memory records, relations, provenance, lifecycle state, and portable workspace data without importing the OpenMinion runtime.

Read This First

  1. Read At a Glance to confirm the package boundary.
  2. Follow Install and Quick Start for a durable SQLite-backed memory round trip.
  3. Read How It Fits before combining SophiaGraph with OpenMinion, PragmaGraph, or GraphFakos.
  4. Use the package docs for workspace, storage, service, retrieval, and production-oriented guidance.
  5. Read Development before changing the package.

Trust and Brand Safety

SophiaGraph has no official token, coin, NFT, airdrop, staking program, treasury product, or investment offering. Any claim otherwise is unauthorized and should be treated as a scam.

At a Glance

Package sophiagraph
Source line v0.0.7 alpha
Python 3.11+
Best fit Durable agent memory, provenance, lifecycle, and local knowledge workspaces
Default durable backend SQLite
Additional backends In-memory, optional Kuzu and Neo4j graph adapters, optional Qdrant vector adapter
Not the claim Agent orchestration, provider routing, or automatic semantic inference

Common Commands

python3.11 -m pip install sophiagraph
sophiagraph-smoke
sophiagraph-ui --screen explore --serve --open

Create and inspect a persistent workspace:

python3.11 -m sophiagraph workspace-init .sophia-workspace \
  --scope agent:local --agent-id local --graph-id main --json
python3.11 -m sophiagraph workspace-status .sophia-workspace --json

Install

Install the base package:

python3.11 -m pip install sophiagraph

Install an optional backend only when you need it:

python3.11 -m pip install "sophiagraph[kuzu]"
python3.11 -m pip install "sophiagraph[neo4j]"

For a source checkout:

python3.11 -m pip install -e ".[dev]"

Quick Start

External Consumer Quickstart

Store and retrieve one typed memory record:

from pathlib import Path
from uuid import uuid4

from sophiagraph.models import MemoryNamespace, MemoryRecord
from sophiagraph.query import SearchQueryOptions
from sophiagraph.storage import create_sqlite_store

store = create_sqlite_store(Path(".sophiagraph"))
namespace = MemoryNamespace(agent_id="demo", graph_id="main")
record = MemoryRecord(
    id=str(uuid4()),
    scope="agent:demo",
    type="fact",
    key="project:apollo",
    title="Apollo launch window",
    content={"text": "Apollo launches in Q2."},
    created_at="2026-01-01T00:00:00+00:00",
    updated_at="2026-01-01T00:00:00+00:00",
    source="validated",
    confidence=0.95,
    namespace=namespace,
)

store.put_record(record)
matches = store.search_records(
    SearchQueryOptions(
        query="Apollo",
        scopes=["agent:demo"],
        namespaces=[namespace],
    )
)
print(matches)

Run the complete example:

python3.11 examples/basic_usage.py

Use create_memory_store() instead when a test or short-lived consumer needs an ephemeral in-memory store rather than a durable SQLite path.

What SophiaGraph Provides

  • typed memory records, namespaces, relations, provenance, and citations
  • SQLite and in-memory stores with explicit capability reporting
  • portable snapshots and import/export contracts
  • lifecycle, trust, governance, deletion, audit, and freshness helpers
  • structural queries, graph paths, connected components, and knowledge views
  • workspace initialization, sync planning, history, templates, and publishing
  • optional graph and vector backend adapters
  • bounded local UI and server surfaces for package-owned workflows

What SophiaGraph Does Not Provide

  • application or agent orchestration
  • model-provider selection or routing
  • session and turn execution
  • automatic fact, relation, tag, or summary inference from prose
  • automatic embedding-provider calls or model selection
  • hosted collaboration, hosted administration, or managed sync
  • ownership of a host application’s authorization or policy decisions

The core package operates on explicit typed inputs. Hosts remain responsible for LLM calls, semantic interpretation, scheduling, credentials, and user-facing policy.

How It Fits

Package Responsibility
OpenMinion Agent runtime, turns, tools, sessions, and orchestration
SophiaGraph Durable memory, provenance, lifecycle, and workspace knowledge
PragmaGraph Deterministic observed facts from code, docs, artifacts, and Git history
GraphFakos Provider-neutral graph viewing and interaction contracts

SophiaGraph can run by itself. OpenMinion may use it as a memory backend, and GraphFakos may visualize a provider projection, but neither integration changes who owns memory truth.

Storage and Workspace Paths

Use the in-memory store for tests and ephemeral consumers. Use SQLite for the default durable local path. Optional Kuzu, Neo4j, and Qdrant adapters expose capability-specific behavior and should be selected explicitly. The Kuzu and Neo4j graph adapters support the same typed structural pattern payloads as the in-memory conformance backend; they never accept natural-language or generated query text.

The workspace surface adds persistent scope, namespace, import, sync, review, and publishing workflows on top of those package contracts:

python3.11 -m sophiagraph workspace-import-plan \
  .sophia-workspace ./notes --json
python3.11 -m sophiagraph workspace-sync-apply \
  .sophia-workspace ./notes --json

Read docs/workspace-mode.md, docs/storage-retrieval-backends.md, and docs/retrieval-boundary.md before building a larger integration.

Delegated Agent Access

Remote agents and untrusted adapters should use AuthorizedSophiaGraphGateway, not a raw store. The gateway intersects a trusted principal context with a host-resolved DelegationMemoryGrant on every operation, bounds results and context, and returns the same absent envelope for missing and unauthorized direct IDs.

Sophiagraph does not persist grants. OpenMinion or another host remains the grant issuer and revocation authority through the package-neutral DelegationMemoryGrantResolver protocol. Child-authored durable knowledge is submitted by the parent as a candidate and follows the existing review and promotion flow.

See docs/delegated-memory-access.md for the trust boundary, migration steps, and rollback posture.

Development

make dev-install
make hooks-install
make check

Use make release-check before publishing or changing the documented public surface.

Docs and Release

License and Brand-use Boundary

  • Source code license: Apache-2.0
  • Brand/trademark grant: none

The license grants rights to use, modify, and redistribute the code. It does not grant rights to present a fork, clone, token, website, or social account as the official SophiaGraph or OpenMinion project or imply affiliation or endorsement.

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

sophiagraph-0.0.7.tar.gz (556.3 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

sophiagraph-0.0.7-py3-none-any.whl (466.8 kB view details)

Uploaded Python 3

File details

Details for the file sophiagraph-0.0.7.tar.gz.

File metadata

  • Download URL: sophiagraph-0.0.7.tar.gz
  • Upload date:
  • Size: 556.3 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/7.0.0 CPython/3.13.14

File hashes

Hashes for sophiagraph-0.0.7.tar.gz
Algorithm Hash digest
SHA256 e94a6d1876f1f02f28ab69feaefaba532980dd3943ca433d6aa78935fd051c8e
MD5 1d835b88799e52df3a1bf4cbb8453836
BLAKE2b-256 b7a8215436931d1c5470ff26383d1f2b7cb42d5fc2f695d2ccc9e3e3ee77ec9e

See more details on using hashes here.

Provenance

The following attestation bundles were made for sophiagraph-0.0.7.tar.gz:

Publisher: release.yml on openminion/sophiagraph

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file sophiagraph-0.0.7-py3-none-any.whl.

File metadata

  • Download URL: sophiagraph-0.0.7-py3-none-any.whl
  • Upload date:
  • Size: 466.8 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/7.0.0 CPython/3.13.14

File hashes

Hashes for sophiagraph-0.0.7-py3-none-any.whl
Algorithm Hash digest
SHA256 4d34047da4d00b2d693c3be61daabe0bb5583e6cbf79ed11cad458e048b14589
MD5 0dc580326b2e961bd869d1aa0b1ab6de
BLAKE2b-256 e321ce92cd0eb644655497b3d03c9f078a48c3d116982f975d3536422a289c3e

See more details on using hashes here.

Provenance

The following attestation bundles were made for sophiagraph-0.0.7-py3-none-any.whl:

Publisher: release.yml on openminion/sophiagraph

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Release history Release notifications | RSS feed

0.0.9

2 files

0.0.8

2 files

This release

0.0.7 This release

2 files

0.0.6

2 files

0.0.5

2 files

0.0.4

2 files

0.0.3

2 files

0.0.2

2 files

0.0.1

2 files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page