Agent Claude Code
Analyzes Lumen benchmark JSONL results to identify chunker/search quality issues and produce actionable improvement recommendations.
254 20d ago A 24 tokens
Save 30% token costs when using Claude Code, Codex, OpenCode for free - with open source, local semantic search. Works for small and large codebases and monorepos! Enterprise-ready and fully compliant via Ollama and SQLite-vec.
Agent Claude Code
Analyzes Lumen benchmark JSONL results to identify chunker/search quality issues and produce actionable improvement recommendations.
Agent Claude Code
Curates bench-swe benchmark tasks from a GitHub issue or PR URL. Requires a URL and language. Extracts commits, generates gold patch, writes task JSON, and verifies inline.