AI Guide V1 — Available Now

Give AI Coding Agents a Map of Your Codebase.

AI Guide is a privacy-first code intelligence layer that helps AI coding agents understand, navigate, and connect the code they need — without sending your source code to our servers.

Get Started View on GitHub

pip install ai-guide-analyzer

pipelinelive

Repositorysource files
AI Guidelocal analysis
Unified Graphentities + relationships
AI Coding Agentstructured navigation

bash — ai-guide

The problem

AI can write code. But first, it needs to find the right code.

Large repositories contain thousands of files, deeply nested modules, dependencies, classes, functions, imports, relationships, and unfamiliar architectures.

AI coding agents can spend valuable context and tool calls navigating the repository before reaching the relevant code.

thousands of files
nested modules
dependencies
classes
functions
imports
relationships
unfamiliar architectures

Large Repository

Too many files

Too much context

Slower / noisier navigation

The solution

AI Guide builds the map. Your AI agent does the thinking.

AI Guide does not replace coding agents — it complements them by supplying structured, evidence-based repository intelligence.

AI Guide

Find
Understand
Connect
Locate
Navigate
Rank

AI Coding Agent

Reason
Plan
Write
Modify
Debug

How it works

From source repository to agent-ready intelligence.

Hover or focus a stage for details.

Unified Graph

One Unified Graph. One source of truth for code navigation.

AI Guide represents repository structure, files, folders, classes, functions, methods, variables, imports, calls, references, relationships, and exact source locations — together in a single graph.

project
 ├── backend
 │    ├── auth.py
 │    │     ├── AuthService
 │    │     └── login()
 │    └── api.py
 │          └── get_users()
 │
 └── main.py

Privacy-first

Your code stays where your code belongs.

AI Guide is designed around local-first code intelligence. The repository and generated code graph can remain inside the developer's environment instead of requiring source-code storage on our servers.

Local-first by designAnalysis and graph generation run in your environment.

YOUR MACHINE

Repository

AI Guide

Code Graph

AI Agent

Deterministic intelligence

No guessing. No black-box repository answers.

AI Guide V1 is parser-based, deterministic, evidence-based, locally executable, and predictable. The deterministic foundation is designed to provide reliable repository structure and relationships to AI agents.

Parser-basedStructure comes from AST/CST parsing.

DeterministicSame repository, same graph.

Evidence-basedRelationships trace back to parsed code.

Locally executableRuns close to the code.

No LLM required
No vector database
No RAG
No natural-language guessing

Query experience

Query the graph from Python, the CLI, or JSON output.

from ai_guide import Guide

guide = Guide("guide-data")

guide.find("AuthService")

guide.where_is("login")

guide.relation(
    "symbol:auth.py:login",
    direction="incoming"
)

guide.path(source, target)

guide analyze ./my-project

guide find AuthService
guide where-is login
guide relation "symbol:auth.py:login"
guide path "symbol:main.py:run_app" "symbol:auth.py:validate_input"

{
  "entity": "symbol:auth.py:login",
  "kind": "function",
  "file": "backend/auth.py",
  "location": { "line": 42, "column": 4 },
  "relationships": [
    { "type": "called_by", "target": "symbol:api.py:get_users" }
  ]
}

Agent integration

Built to work underneath AI coding agents.

AI Guide provides explicit developer-facing interfaces that can be exposed to coding agents. Developers can give their agent integration instructions describing how to use AI Guide's commands and Python API.

Designed for integration with AI coding agents.

Developer

AI Coding Agent

AI Guide

Repository Intelligence

Supported languages

Parser-backed language coverage.

PythonPython AST

JavaScriptTree-sitter

TypeScriptTree-sitter

JSXTree-sitter

TSXTree-sitter

C++Tree-sitter

JavaTree-sitter

Real validation

Already tested on a real codebase.

0Tests Passing

0Entities

0Relationships

0Failed Files

Insight AIReal Repository Test

AI Guide analyzed a real software repository and successfully generated a Unified Graph containing 185 entities and 282 relationships.

Features

What V1 gives your agent.

Unified Graph

Repository structure and code entities represented together.

Exact Source Locations

Line and column metadata attached directly to entities.

Evidence-Based Relationships

Relationships come from repository structure and parser evidence.

Stable Entity IDs

Predictable IDs enable deterministic navigation.

Priority-Ranked Search

Exact matches are prioritized before broader matches.

Local Execution

Designed to operate close to the code.

JSON Output

Machine-readable output for tooling and automation.

Python API

Programmatic access through the Guide interface.

CLI

Simple terminal-first workflow.

Developer workflow

Five steps from install to agent-ready.

Why AI Guide

Search finds text. AI Guide exposes code intelligence.

Generic Code Search

Find matching text

AI Guide

Understand entities
Understand relationships
Navigate structure
Locate exact source
Expose structured intelligence

Roadmap

Where AI Guide is heading.

V1Deterministic Code Intelligence✓
V1.5Developer / IDE Integration→
V2AI Agent Integration→
V2.5Benchmarks + Early Users→
V3AI Code Intelligence Infrastructure→

Items beyond V1 are planned, not yet available.

Vision

From a local developer tool to infrastructure for AI-native software development.

Individual Developers

Development Teams

AI Coding Platforms

Enterprise Infrastructure

AI Navigation Protocol

Our business direction stays privacy-first: intelligence that can run where the code already lives. Pricing is still being finalized and is not published yet.

Give your AI agent a better map.

Start exploring your codebase with deterministic, local-first code intelligence.

Get AI Guide View GitHub

pip install ai-guide-analyzer