Skip to content

CLAUDE.md

Latest commit

 

History

History
58 lines (35 loc) · 1.42 KB

CLAUDE.md

File metadata and controls

58 lines (35 loc) · 1.42 KB
Raw

CLAUDE.md — Local-Bench

Private & Confidential — Property of Lifescope Inc. Do not distribute.

Platform role

Local-Bench sits in the App/Agent layer of the Companion Intelligence platform.

Architecture: CI-Engineering/architecture/architecture.md Roadmap: CI-Engineering/architecture/roadmap.md Issues: https://github.com/companionintelligence/CI-Engineering/issues

What this repo does

Local LLM benchmark and test harness. Evaluates models running via Ollama or CI-Server's LLM facade against defined test cases.

Stack

  • Node.js / JavaScript
  • Jest
  • npm

Key commands

# Install dependencies
npm install

# Start development server / service
# start Ollama or CI-Server first, then:

# Run tests
npm test

# Build for production
# no build — runs directly

Package manager

npm — use exclusively; do not mix with npm/yarn/pnpm/bun unless the repo explicitly requires it.

Cross-repo connections

  • Calls CI-Server /v1/chat/completions (OpenAI-compatible endpoint)
  • Or calls Ollama directly

Running benchmarks

See LLM_TESTS.md and STRIX_HALO.md for specific test suites and hardware notes.

Multi-agent safety

Multiple AI agents work these repos in parallel. Never use git reset --hard, git clean -f, or any destructive git command without explicit confirmation from the user.

Confidentiality

Private & Confidential — Property of Lifescope Inc. Do not distribute.