Zero-token knowledge base

TopoExploit | System Maps

Architecture maps, an overview video, and the full documentation set, generated offline from the source code. Nothing here was written by an LLM.

5communities

44files

414imports

2languages

824symbols

Interactive maps

Physics-driven diagrams: drag nodes, click one for its documentation, trace upstream or downstream reach. Maps load their engine from a CDN.

Force Graph Explorer

Every file, community, layer and external in 2D or 3D: clickable nodes and edges, upstream and downstream reach, path finder, color by community, layer or language, and lenses for import cycles, files nothing imports, and hubs.

107 nodes | 492 edges

Open Force Graph Explorer

Edge Bundle Explorer

Hierarchical edge bundling on a circle or a 3D sphere of community caps: every import routed through its community, thick bundles mark the seams between subsystems, live bundling strength, crossing-only filter.

44 files | 59 wires | 5 communities

Open Edge Bundle Explorer

Runtime Architecture

Components, services, storage, and boundaries across layers.

44 of 44 files in full scope66 links5 chapters

Open architecture map

Data Flow

Movement of data from sources through transforms to stores.

44 of 44 files in full scope66 links5 chapters

Open dataflow map

Change Lifecycle

States, transitions, waits, and retry paths of change.

44 of 44 files in full scope59 links5 chapters

Open lifecycle map

Request Sequence

Ordered interaction between top participants over time.

44 of 44 files in full scope66 links5 chapters

Open sequence map

Delivery Workflow

Delivery path across lanes from entry to terminal outcomes.

44 of 44 files in full scope107 links5 chapters

Open workflow map

Overview video

Cinematic synthwave overview rendered from real scan data: layers, god nodes, dependency tree, communities, import graph, and code DNA. Open/download readmenator.mp4.

Documentation 31 files

Wiki 10

Concept pages per code community: start with index.md.

Second Brain

readmenator-wiki/index.md

The codebase centres on model.py, init.py, train.py. Architecturally it is 6 layers, dominant utility (36 files) across 6 import-based communities. Recorded...

77 lines~736 tokens

Open raw markdown

topogpt3: model

readmenator-wiki/community_0_topogpt3_model.md

Community 0 | 12 files | cohesion 0.64

105 lines~1795 tokens

Open raw markdown

eval: harness

readmenator-wiki/community_1_eval_harness.md

Community 1 | 7 files | cohesion 0.64

93 lines~1789 tokens

Open raw markdown

eval: governor

readmenator-wiki/community_2_eval_governor.md

Community 2 | 6 files | cohesion 0.31

99 lines~1794 tokens

Open raw markdown

topogpt3: inference_hrm

readmenator-wiki/community_3_topogpt3_inference_hrm.md

Community 3 | 5 files | cohesion 0.42

83 lines~1331 tokens

Open raw markdown

topogpt3: jlens

readmenator-wiki/community_4_topogpt3_jlens.md

Community 4 | 4 files | cohesion 0.45

80 lines~1764 tokens

Open raw markdown

orphans

readmenator-wiki/community_5_orphans.md

Community 5 | 10 files | cohesion 0.00

91 lines~1248 tokens

Open raw markdown

Concepts

readmenator-wiki/concepts.md

Second-brain semantic layer: nouns map atomically to file sets (EXTRACTED); verbs aggregate structural edges (INFERRED).

124 lines~2729 tokens

Open raw markdown

Queries

readmenator-wiki/queries.md

Log each answered question here so the wiki compounds. Format: question, answer, cited files.

29 lines~174 tokens

Open raw markdown

Audit Report

readmenator-wiki/REPORT.md

Project: TopoExploit | 2026-10-10 | offline, deterministic

43 lines~361 tokens

Open raw markdown

GraphRAG 1

Community report hierarchy for agents. Query it with: readmenator . ask "question" (or MCP readmenator.graphrag).

GraphRAG Community Reports

readmenator-graphrag/REPORTS.md

Entities: 970 | Relationships: 2759 | Communities: 6 | Themes: 2 | Text units: 860

126 lines~3366 tokens

Open raw markdown

Project docs 3

Knowledge base, README, and policies.

Polyglot Codebase Knowledge Graph

KNOWLEDGE_BASE.md

Start here: Statistics Dashboard for scope, God Nodes for blast radius, Architecture Reference for per-file API. Agents: prefer readmenator-agent/INDEX.md +...

4116 lines~118485 tokens

Open raw markdown

TopoExploit

README.md

A 122M parameter complex-valued autoregressive language model trained on

356 lines~3475 tokens

Open raw markdown

Security Policy

SECURITY.md

Use this section to tell people about which versions of your project are

21 lines~154 tokens

Open raw markdown

Agent docs 12

Grep-friendly indexes for AI agents (also useful for humans).

Index

readmenator-agent/INDEX.md

48 lines~1203 tokens

Open raw markdown

API

readmenator-agent/API.md

Depends on: topogpt3/init.py

598 lines~15719 tokens

Open raw markdown

Architecture

readmenator-agent/ARCHITECTURE.md

app.py -> topogpt3/init.py

105 lines~1818 tokens

Open raw markdown

Concepts

readmenator-agent/CONCEPTS.md

Nouns map atomically to file sets (EXTRACTED); verbs aggregate structural edges (INFERRED).

120 lines~4416 tokens

Open raw markdown

Gotchas

readmenator-agent/GOTCHAS.md

These files have the most connections. Changes here have high blast radius.

48 lines~603 tokens

Open raw markdown

Subsystem: eval

readmenator-agent/KB_eval.md

Doc: Aggregate HumanEval result JSONL files into a summary table.

222 lines~2582 tokens

Open raw markdown

Subsystem: root

readmenator-agent/KB_root.md

Doc: Drop-in entry point that demonstrates how to use the topogpt3 package.

118 lines~1391 tokens

Open raw markdown

Subsystem: tests

readmenator-agent/KB_tests.md

Doc: TestValidPositionMask: Feature: validpositionmask excludes attention-sink and final positions.

100 lines~2181 tokens

Open raw markdown

Subsystem: topogpt3

readmenator-agent/KB_topogpt3.md

Doc: TopoGPT3: complex-valued spectral language model for code.

699 lines~11597 tokens

Open raw markdown

Project Memory

readmenator-agent/MEMORY.md

Generated from 44 files at commit 4d3ec5e49007. Read this first, then readmenator-wiki/index.md, then readmenator . ask "<question>" for anything specific.

84 lines~1227 tokens

Open raw markdown

Security Findings

readmenator-agent/SECURITY.md

3 lines~10 tokens

Open raw markdown

Symbols

readmenator-agent/SYMBOLS.md

837 lines~19704 tokens

Open raw markdown

Recipes 5

Step-by-step tasks grounded in this codebase.

Recipe: Add a Function

readmenator-agent/recipes/add-function.md

Find the target file by purpose: grep -in '<topic>' readmenator-agent/INDEX.md

9 lines~99 tokens

Open raw markdown

Recipe: Change a File Safely

readmenator-agent/recipes/change-impact.md

Riskiest file: topogpt3/continuation.py (26 dependents)

9 lines~109 tokens

Open raw markdown

Recipe: Fix a Dependency Cycle

readmenator-agent/recipes/fix-cycle.md

Read cycles: grep -A5 'Dependency Cycles' readmenator-agent/GOTCHAS.md

6 lines~65 tokens

Open raw markdown

Recipe: Fix a Security Finding

readmenator-agent/recipes/fix-security.md

Read findings: grep -n '<file>' readmenator-agent/SECURITY.md

6 lines~69 tokens

Open raw markdown

Recipe: Reduce File Complexity

readmenator-agent/recipes/reduce-complexity.md

Target hotspot: topogpt3/model.py

9 lines~76 tokens

Open raw markdown

How to read these maps

Open any map, then: drag nodes freely while physics settles the rest, search (/) to filter, click a node to focus it with full file documentation, Upstream and Downstream to trace authored reach, Path to probe the exact route between two ids, Lens to compare semantic roles, Play to walk the guided chapters, Stabilize to re-run physics, Theme for dark and light, Export for PNG or typed JSON. Deep links such as #route=a~b restore any reading.