neuralmind.uk / NeuralMind Use Cases Edit this page on GitHub ↗

NeuralMind Use Cases

Walkthroughs for the most common “what do I actually do?” questions, organized by who you are and what you’re trying to solve. Each page is command-driven — copy, run, done.

Use case Best for Primary goal
Index any repo with just pip (no graphify) First-timers, CI, locked-down machines Nothing → queryable index in one install + one build
Does it work on your code? (5-minute benchmark) Evaluating whether to install at all Measured before/after on YOUR codebase
Claude Code user You use Claude Code daily and want full two-phase optimization Cheapest + smartest agent sessions
Cost optimization Teams or solos watching LLM spend climb Measure, reduce, and report savings — neuralmind savings --cost prices them in dollars (v0.45.0+)
Any LLM (ChatGPT / Gemini / local) You use non-MCP chats or a model-agnostic workflow Get NeuralMind context into any chat window
Offline / regulated work Regulated industries, air-gapped machines 100% local retrieval with zero telemetry
Growing monorepo Codebase where old context goes stale fast Keep the index fresh with minimal effort
Multi-agent codebase You use multiple AI tools (Claude Code + Cursor + Hermes + OpenClaw) on the same project One shared associative memory across every agent; v0.6.0 live graph shows the union
Slim & sovereign: ChromaDB-free local stack Security-sensitive teams, tiny-footprint installs (v0.21.0+) Embed + search with zero ChromaDB — smaller deps, smaller index, fewer advisories
Branch-isolated memory & team baselines Heavy branchers, teams onboarding new devs (v0.24.0+) Keep feature-branch learning out of main’s memory; ship a shared baseline as a versioned bundle
Unified context engineering stack (NeuralMind + Ponytail + Headroom) Teams who’ve hit the ceiling on single-tool optimization Eliminate token waste at retrieval, transport, and generation simultaneously
Blast radius before a rename Anyone (or any agent) about to rename, re-sign, or delete a symbol (v0.42.0+) See every caller / importer / subclass a change would touch, before you edit — from the static code graph
Decision provenance: answer “why is it like this?” Anyone inheriting code whose rationale lives in someone’s head (v0.43.0+) Capture a decision in a Decision: git trailer; recall it with neuralmind why — the why stored where it can’t drift
Find the coverage that lies: mock-only endpoints Anyone with a green suite that still ships DB failures (v0.44.0+) neuralmind gaps classifies endpoints live-covered / mock-only / untested — catch the P2003 before the live smoke
Index a book or docs corpus Anyone pointing NeuralMind at prose rather than code (v3.4.0+) Scope the index to the corpus, skip the code graph, and re-index in seconds as you write — ingest-content --content-only
Measure memory across a major refactor (field report) Anyone rebuilding a subsystem who wants proof the memory adapts Before/after snapshot recipe + real case study: 48.8× reduction, personal synapse edges 36→135 on a ~9.3k-node TypeScript SaaS platform

Not sure which applies? Start with the symptom / goal table in the main README.