npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

axiongraph

v0.0.5

Published

Invisible events. Replayable graphs. AxionGraph is an append-only execution-graph event model and deterministic reducer — provider-agnostic, storage is a port.

Readme

AxionGraph

Invisible events. Replayable graphs.

AxionGraph is an append-only event model and deterministic reducer for execution graphs. It records graph events from agents, tools, workflows, and connectors, then folds them into portable graph state for storage, replay, testing, and visualization.

Storage is a port. Rendering is a consumer concern. The core knows nothing about Convex, Neo4j, realtime transports, auth, policy, or budgets — it is a small, byte-stable, parity-testable primitive that ships as MIT OSS.

import { reduceAll } from "axiongraph";

const events = [
  { id: "evt_01", runId: "run_01", seq: 1, ts: "2026-06-02T12:00:00.000Z",
    type: "node_created", node: { id: "agent_research", kind: "agent", label: "Research Agent" } },
  { id: "evt_02", runId: "run_01", seq: 2, ts: "2026-06-02T12:00:01.000Z",
    type: "node_created", node: { id: "tool_web", kind: "tool", label: "Web Search" } },
  { id: "evt_03", runId: "run_01", seq: 3, ts: "2026-06-02T12:00:02.000Z",
    type: "edge_created", edge: { id: "edge_01", kind: "called_tool", from: "agent_research", to: "tool_web", status: "completed" } },
];

const state = reduceAll("run_01", events);
console.log(state.nodes.size); // 2
console.log(state.edges.size); // 1

Why

Agent and workflow systems need a clean way to answer:

  • What happened, and in what order?
  • Which agents, tools, sources, and artifacts were involved?
  • Can the run be replayed? Can the graph state be tested across runtimes?

AxionGraph keeps that layer small and portable.

Core ideas

  • Append-only events are the source of truth; graph state is derived by folding them.
  • The reducer is pure and deterministic — identical event logs fold to byte-identical state.
  • A monotonic seq per run defines order; wall-clock ts is advisory.
  • Node/edge kind is an open taxonomy; supply a GraphVocabulary to reject unknown kinds.
  • Storage is a port (GraphStore); rendering and realtime transport are consumer concerns.
  • TypeScript first, with a Python mirror kept honest by shared parity fixtures.

One package, opt-in extras

AxionGraph ships as a single axiongraph package with subpath entry points — the npm equivalent of Python extras. Heavy backends are declared as optional peer dependencies, so you install only what a feature needs.

| Entry point | Description | Extra to install | | --- | --- | --- | | axiongraph | Event model, deterministic reducer, canonicalizer, vocabulary machinery, and the GraphStore port. | — | | axiongraph/store-local | Zero-service reference adapters: an in-memory store and a node:sqlite-backed durable store. | — (node:sqlite is built in) | | axiongraph/store-postgres | Durable PostgresStore backed by a pg pool: jsonb event log keyed on (runId, seq), idempotent appends, live-fold snapshots. | pg |

Planned subpaths: axiongraph/store-convex (peer: convex), axiongraph/store-neo4j (peer: neo4j-driver). A Python mirror ships the same adapters as PyPI extras.

Install

pnpm add axiongraph
# later, opting into a backend extra, e.g.:
# pnpm add axiongraph convex

Storing and replaying events

import { SqliteStore } from "axiongraph/store-local";

const store = new SqliteStore("./run.db"); // or new InMemoryStore()
await store.append(events);                // idempotent on (runId, seq)
const state = await store.snapshot("run_01");

Both InMemoryStore and SqliteStore satisfy the same GraphStore contract, so they are interchangeable; any future adapter that passes the shared contract suite drops in the same way.

Development

Node 24 and pnpm 9. The repo is a pnpm workspace.

pnpm install
pnpm verify   # biome check + tsc typecheck + Vitest
pnpm build    # bundle the internal packages into the single axiongraph dist

The repo is an internal pnpm workspace (packages/core, packages/store-local, packages/store-postgres, plus a dev-only packages/testkit shared contract suite); tsup bundles the publishable ones into the single axiongraph dist with subpath exports.

The Postgres contract suite is gated on AXIONGRAPH_TEST_POSTGRES_URL; it is skipped locally unless set, and CI runs it against a postgres:16 service.

Status

Early development. The TypeScript core and reference stores are implemented and parity-tested; the Python mirror follows.

License

MIT