Codebase intelligence.For people and AI.
SRMYour code.
In perspective.
Explore your code visually. Understand connections across repositories. Bring relevant context to your AI tools. Locally, on your machine.
Personal walkthrough · No code upload required

01 / EXPLORE
What connects
to this code?
Before a change, explore the relationships around it. Follow dependencies from the bigger picture to the relevant source.
CITY / SEE THE WHOLE
Where should
you look closer?
See the shape of a codebase and explore areas worth a closer look. Give architecture discussions a shared visual reference.
ATLAS / GET YOUR BEARINGS
New to the code?
Start with a map.
An unfamiliar repository is easier to discuss when everyone has a point of reference. Explore its regions in Atlas, then follow a question into the code.
For onboarding, architecture reviews and the next person joining the conversation.
GALAXY / ACROSS REPOSITORIES
Your work spans repos.
Your view can too.
When one repository is only part of the picture, Galaxy brings multiple mapped repositories into a shared portfolio view.
Keep the wider picture.
Review several codebases in one place.
Keep the detail within reach.
Move from the portfolio to an individual repository.
Share a point of reference.
Bring a multi-repository view to technical conversations.
02 / REAL CODEBASES
Big code.
No server to manage.
Linux and Chromium, mapped and explored on a laptop. SRM runs locally; no separate SRM server infrastructure is required.


Real, navigable maps. Captured on a MacBook Pro with Apple M4 Pro and 24 GiB of memory. Counts describe the captured repositories.
03 / WORKING WITH CODE
Less time finding.
More context to act.
Find a starting point.
Explore unfamiliar structure before diving into files.
Follow a question.
Trace a connection and reach the relevant source.
Review together.
Use the same map to discuss a change with your team.
04 / YOUR AI WORKFLOW
Better context.
Less searching blind.
Help your agent find relevant code and relationships before proposing a change. Bring SRM context to the tools you already use through MCP.
Gemini · Claude · Codex
And other compatible MCP clients.
START WITH A QUESTION
- Where does this feature live?
- Which connections should I examine before changing it?
- What code does my AI need for this task?
Explore structure, relationships and source through MCP.
MEASURED / TASK-SPECIFIC
Fewer tokens.
More accurate answers.
MoneyChess · Luna · 25 paired comparisons using supplied SRM context. Results depend on the task and context delivery.
Total tokens across four semantic task classes, compared with shell-only.
Mean answer accuracy across all five task classes: SRM context vs shell-only.
Study scope & limits
MoneyChess repository · Luna model · five tasks, five paired runs per task (25 pairs). The token result covers the four semantic classes; accuracy covers all five. The study supplied SRM context before the model answered. It did not test autonomous MCP tool selection or establish these gains for every model, client or repository. Token counts are not a universal cost-saving percentage.
05 / LOCAL BY DESIGN
Your code stays local.
You choose what to share.
SRM maps repositories on your machine. External AI services receive only what you choose to share through your tools and their configuration.
A personal walkthrough with the creator. See Linux, Chromium and the views that matter to your team. No code upload or repository access required.