Neuronpedia is an open-source platform for examining how neural networks work through model activations, circuits, feature representations, and related analyses. Researchers and developers use it to inspect, search, score, visualize, and export interpretability data. Its catalogue add-ons support local development and workflows around the platform.
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.
npx agentmods add instructions/hijohnnylin/neuronpedia/agents-mdgit clone --depth 1 https://github.com/hijohnnylin/neuronpediaWrote 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.
[](https://agentmods.dev/instructions/hijohnnylin/neuronpedia/agents-md)<a href="https://agentmods.dev/instructions/hijohnnylin/neuronpedia/agents-md"><img src="https://agentmods.dev/badge/instructions/hijohnnylin/neuronpedia/agents-md.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.07202 | $0.07202 |
| Opus 5 | $0.03601 | $0.03601 |
| Sonnet 5 | $0.01440 | $0.01440 |
| Haiku 4.5 | $0.00720 | $0.00720 |
Grade A, and why
neuronpedia AGENTS.md 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 4d 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.
How it starts
The opening of the file, as written. The whole thing — 466 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Neuronpedia Development Guide
This file is the single source of truth for agent instructions, and every harness reads it — see "Agent instruction files" at the bottom before adding rules anywhere else.
Repository layout
apps/
autointerp/
graph/
inference/
nla/
sparsity/
webapp/
lib/api/ # generated: <app>.d.ts from each server's openapi.json
docs/ # the long-form reasoning this file's rules point at
utils/
webapp-python-client/ # hand-written public SDK
Neuronpedia is an interpretability website/platform: a reference for what is inside a neural network, plus tools for researchers to experiment with those internals — viewing activations for a given 'neuron' (or feature/latent), steering on them, tracing circuits.
apps/ holds most of the code. Five of the six are Python FastAPI servers, called by the webapp
rather than by users directly:
- inference has the actual model loaded, and serves anything requiring a forward pass. The
webapp calls it mostly from
apps/webapp/lib/utils/inference.ts. - autointerp explains and scores neurons/features/latents. Called mostly from
apps/webapp/lib/utils/autointerp.ts. - graph, nla and sparsity are the attribution-graph, Natural Language Autoencoder (activation vector <-> text) and sparse-circuit services.
webapp is the sixth: a Next.js app that is both the frontend and the user-facing API under
apps/webapp/app/api. Its database schema is apps/webapp/prisma/schema.prisma, and it has its
own AGENTS.md with frontend conventions.
Cross-server types are generated from the Python, in one direction. Each server's pydantic models
are the source of truth; make openapi regenerates every spec and every lib/api/<app>.d.ts from
them. There is no hand-written spec and no client package in the loop, so adding an endpoint means
writing pydantic models, not editing a schema file. The full workflow, and which servers are
camelCase on the wire versus deliberately snake_case, is under "Cross-server APIs" below.
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.
- 4d ago First seen · 466 lines · 7,202 tokens per session scan A 5c4a70ee65c5
neuronpedia AGENTS.md is an instructions file published in the GitHub repository hijohnnylin/neuronpedia (1,122 stars, last pushed yesterday), licensed Apache-2.0. It adds 7,202 tokens to every session, about $0.0360 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 2026-08-30.
Other instructions, from other repositories
nnsight CLAUDE.md
Instructions for ndif-team/nnsight, covering nnsight — agent guide, how to use this file, by task, "i want multi-token / autoregressive generation" and "i want to run multiple prompts at once".
ndif CLAUDE.md
Claude Code instructions for ndif-team/ndif, covering claude.md — ndif agent guide, what ndif is, top-level layout, architecture at a glance and security sandbox (the highest-stakes area).
agent-lens CLAUDE.md
Instructions for dreadnode/agent-lens, covering agentlens, project structure, running experiments, config format (yaml) and engines.
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
spec-kit AGENTS.md
AGENTS.md instructions for github/spec-kit, covering agents.md, about spec kit and specify, quickstart — add a new integration in 5 steps, integration architecture and integrationmanifest — file tracking.
codex AGENTS.md
AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.