kaos-graph
Part of Kelvin Agentic OS (KAOS) — open agentic infrastructure for legal work, built by 273 Ventures. See the full KAOS package map for the rest of the stack.
kaos-graph is a high-performance graph library for KAOS — a typed Python
API over a pure-Rust core backed by
petgraph and the
oxrdf RDF engine. It provides the
graph-algorithm and knowledge-graph building blocks the rest of the stack
relies on: 40+ algorithms (PageRank, shortest paths, centrality, community
detection, DAG ops), RDF/SPARQL parsing, and typed Python wrappers
throughout.
It is dependency-light: the BASE install pulls only kaos-graph itself
plus defusedxml (~250 KB pure-Python). Optional extras layer in the
rest of the KAOS ecosystem.
Install
uv add "kaos-graph>=0.1.0"
# or
pip install "kaos-graph>=0.1.0"
kaos-graph requires Python 3.13 or newer. The published wheels
are cp313-abi3 — one wheel per OS/architecture covers every CPython
3.13+ minor (3.13, 3.14, 3.15, …). No re-release needed when 3.15 ships.
Platform coverage: Linux x86_64 (manylinux + musllinux), Linux aarch64 (manylinux + musllinux), macOS arm64, Windows x86_64, Windows arm64.
Quick start
from kaos_graph import Graph
from kaos_graph.algorithms import pagerank, shortest_paths, topological_sort
g = Graph()
g.add_node("alice", role="engineer")
g.add_node("bob", role="manager")
g.add_edge("alice", "bob", relationship="reports_to")
# Algorithms
ranks = pagerank(g)
costs = shortest_paths(g, "alice")
order = topological_sort(g)
Concepts
The package is built around a small set of typed primitives.
| Concept | What it is |
|---|---|
Graph |
Property graph backed by petgraph::StableDiGraph. String-keyed nodes with JSON-valued properties; edges with arbitrary properties; multi-graph semantics opt-in. Stable node refs across mutations (additions/removals don't invalidate handles). |
Node / Edge / Triple |
Typed dataclass results emitted at the FFI boundary. Node(id, properties), Edge(source, target, properties), Triple(subject, predicate, object). |
| Algorithms | 40+ functions in kaos_graph.algorithms covering traversal (BFS/DFS, topological sort), shortest paths (Dijkstra, A*, Bellman-Ford, Floyd-Warshall), centrality (PageRank, betweenness, closeness, eigenvector), community (Louvain, label propagation), connectivity (SCC, weakly-connected, articulation points, bridges), structure (cycles, cliques, isomorphism, max flow), and DAG ops (longest path, critical path). |
| Sparse PageRank | O((V+E)·n) instead of petgraph's O(V²·E·n). FOLIO ontology (~25K nodes, ~36K edges): 7 ms vs petgraph's 105 s. The marketing-grade benchmark exercised on every CI run. |
| RDF / SPARQL | oxrdf + oxrdfio parsers (Turtle, N-Triples, N-Quads, RDF/XML, TriG) with byte / triple / size caps applied at the FFI boundary. SPARQL via pyoxigraph opt-in through the [rdf] extra. |
| I/O & bridges | JSON (round-trippable), GraphML / GEXF (defusedxml-hardened), adjacency-list, mermaid, dot. kaos-content TabularDocument bridge available via the [tabular] extra. |
| Schema | GraphSchema, NodeType, EdgeType, SchemaViolation for property-graph validation. |
| Storage | VFS-backed save_to_vfs / load_from_vfs with name validation against namespace escape. |
CLI
kaos-graph ships a kaos-graph administrative CLI plus an optional
kaos-graph-serve MCP server (loopback-only by default):
kaos-graph version # version info
kaos-graph serve # forwards to kaos-graph-serve (stdio MCP)
kaos-graph-serve # MCP server, stdio transport
kaos-graph-serve --http # MCP server, streamable HTTP (loopback only)
Note: 17 MCP tools are registered by register_graph_tools() and are
available with pip install "kaos-graph[mcp]>=0.1.0" (pulls in
kaos-mcp).
Compatibility & status
| Aspect | |
|---|---|
| Python | 3.13, 3.14 (informational matrix entries for 3.14t free-threaded and 3.15-dev). One cp313-abi3 wheel per OS/arch covers all 3.13+ minors. |
| OS | Linux (manylinux + musllinux, x86_64 + aarch64), macOS arm64, Windows x86_64, Windows arm64. macOS x86_64 deliberately skipped (Apple ended Intel sales in 2023). |
| Maturity | 0.1.0 GA. The top-level kaos_graph.__all__ documents the core types (Graph, Node, Edge, Triple, BfsNode, DfsEvent, __version__). The full public API is layered by subpackage and is documented in each subpackage's own __all__: kaos_graph.algorithms.__all__ (42 algorithm functions), kaos_graph.schema.__all__ (GraphSchema, NodeType, EdgeType, SchemaViolation), kaos_graph.io.__all__ (JSON, GraphML, GEXF, adjacency-list, mermaid, dot), kaos_graph.rdf.__all__ (RDF/SPARQL via the [rdf] extra), kaos_graph.knowledge.__all__, kaos_graph.tools.__all__ (MCP tools), and kaos_graph.errors.__all__. Import from the subpackage that owns the symbol (from kaos_graph.algorithms import pagerank, etc.). |
| Stability policy | Pre-1.0: minor bumps may change behaviour. Every change is documented in CHANGELOG.md. |
| Test coverage | 173 Rust unit tests + 402 Python tests passing (575 total). 38 dedicated security regressions covering the audit-pass A2 + two follow-up rounds. |
| Type checker | Validated with ty, Astral's Python type checker. |
Documentation
Per-package reference: docs/ in this repo.
Cross-cutting KAOS guides (agentic patterns, persona presets, settings
policy, citations, MCP data flow, migration to 0.1.0 GA) live in
kaos-modules/docs/guides/.
Companion packages
kaos-graph is one of the packages in the
Kelvin Agentic OS. The broader stack:
| Package | Layer | What it does |
|---|---|---|
kaos-core |
Core | Foundational runtime, MCP-native types, registries, execution engine, VFS |
kaos-content |
Core | Typed document AST: Block/Inline, provenance, views |
kaos-mcp |
Bridge | FastMCP server, kaos management CLI, MCP resource templates |
kaos-pdf |
Extraction | PDF → AST with provenance |
kaos-web |
Extraction | Web extraction, browser automation, search, domain intelligence |
kaos-office |
Extraction | DOCX / PPTX / XLSX readers + writers to AST |
kaos-tabular |
Extraction | DuckDB-powered SQL analytics |
kaos-source |
Data | Government + financial data connectors (Federal Register, eCFR, EDGAR, GovInfo, PACER, GLEIF) |
kaos-llm-client |
LLM | Multi-provider LLM transport |
kaos-llm-core |
LLM | Typed LLM programming (Signatures, Programs, Optimizers) |
kaos-nlp-core |
Primitives (Rust) | High-performance NLP primitives |
kaos-nlp-transformers |
ML | Dense embeddings + retrieval |
kaos-graph |
Primitives (Rust) | Graph algorithms + RDF/SPARQL |
kaos-ml-core |
Primitives (Rust) | Classical ML on the document AST |
kaos-citations |
Legal | Legal citation extraction, resolution, verification |
kaos-agents |
Agentic | Agent runtime, memory, recipes |
kaos-reference |
Sample | Reference module for module authors |
Packages depend on kaos-core; everything else is opt-in. Mix and match the
ones you need.
Development
git clone https://github.com/273v/kaos-graph
cd kaos-graph
uv sync --group dev --extra rdf
uv run maturin develop --release
Install pre-commit hooks (recommended — they run the same checks as CI on every commit, scoped to staged files):
uvx pre-commit install
uvx pre-commit run --all-files # one-time full sweep
Manual QA commands (the same set CI runs):
cargo fmt --check
cargo clippy --no-default-features --all-targets -- -D warnings
cargo test --no-default-features --lib
uv run ruff format --check python/kaos_graph tests
uv run ruff check python/kaos_graph tests
uv run ty check python/kaos_graph tests
uv run pytest tests/
Build from source
uv build
uv pip install dist/*.whl
Contributing
Issues and pull requests are welcome. See CONTRIBUTING.md
for setup, quality gates, pull request expectations, and engineering
standards. By contributing you agree to follow the
project conduct expectations and certify the
Developer Certificate of Origin v1.1 —
sign every commit with git commit -s. Please open an issue before starting
on a non-trivial change so we can align on scope.
Security
For security issues, please do not file a public issue. Report privately via GitHub Private Vulnerability Reporting or email security@273ventures.com. See SECURITY.md for the full disclosure policy.
License
Apache License 2.0 — see LICENSE and NOTICE.
Copyright 2026 273 Ventures LLC. Built for kelvin.legal.
Release files for kaos-graph 0.1.5
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| kaos_graph-0.1.5.tar.gz | 268.1 kB | Details |
Built distributions (wheels)
| File | Reset | |||
|---|---|---|---|---|
| kaos_graph-0.1.5-cp313-abi3-win_arm64.whl | CPython 3.13 | abi3 | Windows ARM64 | Details |
| kaos_graph-0.1.5-cp313-abi3-win_amd64.whl | CPython 3.13 | abi3 | Windows x86-64 | Details |
| kaos_graph-0.1.5-cp313-abi3-manylinux_2_28_x86_64.whl | CPython 3.13 | abi3 | Linux glibc 2.28+ x86-64 | Details |
| kaos_graph-0.1.5-cp313-abi3-manylinux_2_28_aarch64.whl | CPython 3.13 | abi3 | Linux glibc 2.28+ ARM64 | Details |
| kaos_graph-0.1.5-cp313-abi3-macosx_11_0_arm64.whl | CPython 3.13 | abi3 | macOS 11.0+ ARM64 | Details |
Total release size: 5.4 MB
Release files / kaos_graph-0.1.5.tar.gz
| Download URL | kaos_graph-0.1.5.tar.gz |
|---|---|
| Size | 268.1 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
7e302fe5a7e56459680f473771739dbbac859042422a41e117d33ee565eccd55
|
|
BLAKE2b-256 checksum How to use checksums |
60eab4ede9ba84e5a2e93284c5184916c6b876145f67c69b4a0fe64130199abb
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 23, 2026.
Transparency logRelease files / kaos_graph-0.1.5-cp313-abi3-win_arm64.whl
| Download URL | kaos_graph-0.1.5-cp313-abi3-win_arm64.whl |
|---|---|
| Size | 960.9 kB |
| Tags | CPython 3.13 Windows ARM64 abi3 |
|
SHA-256 checksum How to use checksums |
c786d803af950a3c03e95352d4f781e707b8054743fb2029b7e4fa7c1aa49f96
|
|
BLAKE2b-256 checksum How to use checksums |
aeba9ff2887f4839f54c8b31debbfd6e0dc735018738c50a0d0e2d3f1a786774
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 23, 2026.
Transparency logRelease files / kaos_graph-0.1.5-cp313-abi3-win_amd64.whl
| Download URL | kaos_graph-0.1.5-cp313-abi3-win_amd64.whl |
|---|---|
| Size | 1.0 MB |
| Tags | CPython 3.13 Windows x86-64 abi3 |
|
SHA-256 checksum How to use checksums |
8caf3b2163e19d78fc107c3b0667c14b8a906aedafe8f99c5890806274cc8eef
|
|
BLAKE2b-256 checksum How to use checksums |
e8e2de6631186a5209a663834fc6adc5572efff7ce11a3b7476a05ed37c6f62c
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 23, 2026.
Transparency logRelease files / kaos_graph-0.1.5-cp313-abi3-manylinux_2_28_x86_64.whl
| Download URL | kaos_graph-0.1.5-cp313-abi3-manylinux_2_28_x86_64.whl |
|---|---|
| Size | 1.1 MB |
| Tags | CPython 3.13 Linux glibc 2.28+ x86-64 abi3 |
|
SHA-256 checksum How to use checksums |
bffda4041fa47a44cf28696ba8256ddede0324bbfa02dada6efc2b654bb602a7
|
|
BLAKE2b-256 checksum How to use checksums |
69bf568e80b223b5808a69fd74a5a1894221342a10e6d29e62c8f2fc09e3fe7c
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 23, 2026.
Transparency logRelease files / kaos_graph-0.1.5-cp313-abi3-manylinux_2_28_aarch64.whl
| Download URL | kaos_graph-0.1.5-cp313-abi3-manylinux_2_28_aarch64.whl |
|---|---|
| Size | 1.0 MB |
| Tags | CPython 3.13 Linux glibc 2.28+ ARM64 abi3 |
|
SHA-256 checksum How to use checksums |
d403435e483da3365566aeafedab7808a6bb5225dbfe262eab1b6eeee5b33557
|
|
BLAKE2b-256 checksum How to use checksums |
b416e528a7d80180e6ca08cc1682440536054dbba8ea7f52bf3ddce76c79a319
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 23, 2026.
Transparency logRelease files / kaos_graph-0.1.5-cp313-abi3-macosx_11_0_arm64.whl
| Download URL | kaos_graph-0.1.5-cp313-abi3-macosx_11_0_arm64.whl |
|---|---|
| Size | 1.0 MB |
| Tags | CPython 3.13 abi3 macOS 11.0+ ARM64 |
|
SHA-256 checksum How to use checksums |
027f3b243e174986179b68f6fd7fa6a517a9be0684de6c3be48496404b7ed001
|
|
BLAKE2b-256 checksum How to use checksums |
f395d4a7c76b52fbb6c6cff14e31e1a2504792f6db1e5a92309ef732e522bed9
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 23, 2026.
Transparency log