This document gives SBE a practical open-source launch plan and ready-to-post copy.
For the current campaign sequence, use docs/launch-campaign.md. For long-form article drafts, use docs/articles/medium-sbe-launch.md and docs/articles/daily-dev-sbe-launch.md.
Software Brain Engine is a local Rust CLI that helps AI coding tools avoid reading unnecessary code. It indexes TypeScript/TSX repositories, finds impacted symbols/files/layers for a planned change, and reports approximate token savings.
Short version:
SBE is a local code-intelligence CLI that gives LLMs focused change context instead of the full repo.
One-line pitch:
I built an open-source Rust CLI that scans TypeScript repos and shows what code an LLM actually needs for a change, with token-savings benchmarks.
Problem:
AI coding tools often burn tokens exploring broad folders before they understand a targeted change.
Solution:
SBE builds a local .sbe index, maps symbols/imports/references, classifies impacted layers, and returns a benchmarkable context packet.
Best proof:
sbe benchmark C:\repo --query "jwt to passport"
Download page:
https://github.com/sarathkumar1207/software-brain-engine/releases/latest
Before posting widely:
HN usually responds better to honest technical framing than polished marketing.
Title:
Show HN: SBE, a Rust CLI that indexes TypeScript repos for LLM context
Post text:
I built Software Brain Engine, an open-source Rust CLI for TypeScript/TSX repositories.
The idea is to reduce wasted LLM context. Instead of sending broad folders or the whole repo, SBE scans locally, stores a binary .sbe index, extracts symbols/imports/references, and answers questions like:
sbe benchmark ./repo --query "jwt to passport"
It reports matched symbols, impacted files, impacted layers such as Auth/Middleware/Controller/DTO/Database, query time, and approximate full-context vs focused-context tokens.
This is production-alpha, not a full TypeScript type checker. Current analysis is syntax-based through Tree-sitter. I am sharing it early because I want feedback on the benchmark methodology, resolver accuracy, and whether this is useful for AI coding workflows.
Repo: <GitHub URL>
Website: <Website URL>
First comment:
Important limitation: the token estimator is currently approximate, based on source characters. It is meant to compare full-project context vs focused SBE context, not to be model-tokenizer exact yet.
Also, small projects can show 0% savings because metadata overhead is larger than the focused code slice. That is intentional: the benchmark should show misses honestly.
daily.dev works well with a practical developer-tool angle.
Title:
I built SBE: a local Rust CLI to reduce LLM code-context tokens
Post:
AI coding tools are useful, but they often inspect too much code before understanding a focused change.
Software Brain Engine is an open-source Rust CLI that indexes TypeScript/TSX repos locally and answers change-impact questions:
sbe benchmark ./repo --query "jwt to passport"
It returns:
- matched symbols
- impacted files
- impacted layers
- dependency/dependent context
- approximate full-repo vs focused-context tokens
The goal is not magic compression. The goal is benchmarkable context selection before an LLM call.
It is production-alpha: local binary index, CLI UX, validation reports, Windows MSI workflow, and tests. Type-aware resolution and exact tokenizer support are next.
GitHub: <GitHub URL>
Use a clearer business/developer productivity framing.
I am building Software Brain Engine, an open-source Rust CLI for AI-assisted software development.
The problem: LLM coding workflows often waste tokens by reading too much of the codebase before understanding a specific change.
SBE indexes a TypeScript repository locally and answers questions like:
"What changes if we migrate JWT auth to Passport?"
It returns impacted symbols, files, dependency paths, code layers, and an approximate token-savings benchmark.
The goal is simple: give AI tools the code that matters, not the entire repository.
Current alpha includes:
- TypeScript/TSX scanning
- local .sbe binary index
- impact and graph queries
- benchmark and validation reports
- Windows MSI release workflow
I am looking for feedback from developers using AI coding tools on real TypeScript backends.
GitHub: <GitHub URL>
Thread:
1/ I am building SBE: an open-source Rust CLI that helps LLM coding tools avoid reading unnecessary code.
2/ Run:
sbe benchmark ./repo --query "jwt to passport"
SBE scans TypeScript/TSX locally and reports impacted symbols, files, layers, dependencies, and token estimates.
3/ Why? AI coding tools often burn context exploring broad folders. SBE creates a local .sbe index so agents can ask for focused code context first.
4/ It is production-alpha: binary storage, CLI UX, validation reports, Windows MSI workflow, tests, and docs.
5/ It is honest about limits: syntax-based analysis for now, approximate token estimate, exact tokenizer and richer resolver coming next.
GitHub: <GitHub URL>
Use only communities that allow project sharing. Keep the post technical and ask for feedback.
Title:
Feedback wanted: Rust CLI for reducing LLM context in TypeScript repos
Post:
I am working on an open-source tool called Software Brain Engine.
It scans a TypeScript/TSX repo locally, builds a .sbe index, and answers planned-change queries such as "jwt to passport". The output includes impacted files, symbols, layers, dependencies, and approximate full-repo vs focused-context tokens.
I am not trying to claim perfect static analysis. V1 is syntax-based with Tree-sitter. The goal is to validate whether a local impact index can reduce LLM context before sending code to an AI coding tool.
I would appreciate feedback on:
- benchmark methodology
- TypeScript resolver accuracy
- CLI UX
- whether this would help your AI coding workflow
Repo: <GitHub URL>
Wait until the release artifacts and website are stable. Product Hunt-style launches need a polished demo and screenshots.
Suggested tagline:
Local code intelligence for smaller, smarter LLM context.
Description:
Software Brain Engine indexes TypeScript repositories locally and returns focused impact reports for planned code changes, helping AI coding tools reduce unnecessary token usage.
Add badges only after workflows are stable:
[](https://github.com/<owner>/<repo>/actions/workflows/rust.yml)
[](https://github.com/<owner>/<repo>/actions/workflows/release.yml)
[](LICENSE)
Avoid these claims until the project proves them:
Use this instead:
SBE is a production-alpha tool for benchmarkable, local-first context selection.