graph_representation__graphdef___graphstate_

graph_representation__graphdef___graphstate_ is a cursor rule for coding agents from altaidevorg/rules-for-ai. It costs 28 tokens per session (4,095 once invoked), scanned A, original, MIT.

The internal representation of a Flax NNX neural-network module, split into its fixed structure and changing values. GraphDef describes the layers and connections, while GraphState contains the current data.

In plain words
What is it for?
Use it to understand or manage how NNX represents nested modules, shared objects, and mutable values when connecting them to JAX.
Why use it?
JAX needs model structure and runtime values handled differently. Separating them makes stateful NNX modules usable with JAX’s transformations.

Cursor rule

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add rules/altaidevorg/rules-for-ai/graph_representation__graphdef___graphstate_
Clone the repo
git clone --depth 1 https://github.com/altaidevorg/rules-for-ai

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for graph_representation__graphdef___graphstate_

README.md
[![agentmods](https://agentmods.dev/badge/rules/altaidevorg/rules-for-ai/graph_representation__graphdef___graphstate_.svg)](https://agentmods.dev/rules/altaidevorg/rules-for-ai/graph_representation__graphdef___graphstate_)
Your own site
<a href="https://agentmods.dev/rules/altaidevorg/rules-for-ai/graph_representation__graphdef___graphstate_"><img src="https://agentmods.dev/badge/rules/altaidevorg/rules-for-ai/graph_representation__graphdef___graphstate_.svg" alt="Measured on agentmods" height="20"></a>
Per session 28 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 4,095 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00028 $0.04095
Opus 5 $0.00014 $0.02048
Sonnet 5 $0.00006 $0.00819
Haiku 4.5 $0.00003 $0.00409

Measured 3d ago against content hash 6c290acf6990, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

graph_representation__graphdef___graphstate_ scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

examples/flax/graph_representation__graphdef___graphstate_.mdc · 244 lines

How it starts

The opening of the file, as written. The whole thing — 244 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Chapter 7: Graph Representation (GraphDef / GraphState)

In the previous chapter, we explored the NNX Functional API (nnx.split, nnx.merge, etc.) which provides the bridge between stateful nnx.Module objects and JAX's functional world. This chapter delves into the core data structures returned and consumed by this API: GraphDef and GraphState. These structures represent the internal state of an nnx.Module graph in a way that is compatible with JAX transformations.

Motivation: Separating Structure from State for JAX

JAX requires pure functions and immutable pytrees for its transformations (jax.jit, jax.grad, etc.). However, nnx.Modules are inherently stateful Python objects. To make these work with JAX, NNX needs a way to represent them functionally. The solution is to split the nnx.Module into two distinct parts:

  1. Static Structure: Everything about the module's composition that doesn't change during typical execution – the types of layers, their attribute names, the nesting hierarchy, and importantly, how different parts might share underlying objects (like shared weights). This needs to be static from JAX's perspective, ideally hashable, so it can be part of the compilation cache key.
  2. Dynamic State: The actual values held within the module, primarily the JAX arrays constituting parameters, batch statistics, optimizer states, etc. This is the part that changes during training or inference and needs to be processed by JAX transformations. It must be represented as a JAX pytree.

GraphDef captures the static structure, and GraphState holds the dynamic state. The functional API functions (nnx.split, nnx.merge) are responsible for converting between the stateful nnx.Module object and this (GraphDef, GraphState) pair.

Central Use Case: Inspecting split Output

The most direct way to interact with GraphDef and GraphState is by calling nnx.split. Let's see what it produces:

Read the full file on GitHub · 244 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 3d ago First seen · 244 lines · 28 tokens per session scan A 6c290acf6990

Subscribe to this mod's changes

graph_representation__graphdef___graphstate_ is a cursor rule published in the GitHub repository altaidevorg/rules-for-ai (2 stars, last pushed 1y ago), licensed MIT. It adds 28 tokens to every session and 4,095 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.