Build a Code Graph Before Agents Read a Single File, Says Limestone
alex_verem · x · 2026-10-15
LimestoneHQ shares a workflow to make coding agents far more efficient: instead of rereading the codebase on every ticket, pre-build a code graph where files, functions, classes and types become nodes and calls/imports/references become edges, clustered into modules via community detection.
Key details: fix the seed and pin tool versions for reproducible graphs; keep AI out of the build (use a model only to name modules); map code only, not docs; the full build takes 16 seconds. Then add agent rules: query the graph before reading any source, ask "what breaks if I change this?" before every change, and rebuild after each merge to main.
Related event: Code graphs cut coding agent token costs by ~25%(2 posts)→
More from coding & agent
- sgrep cuts coding agent LLM costs 21.3% on SWE-bench versus ripgrep, fully local in Rust — udmrzn · 2026-10-15
- 10 GitHub repos that turn Claude Code and Codex into elite UI designers — udmrzn · 2026-10-15
- Grep adds 1,000+ Composio integrations and authenticated browser access for agents — ycombinator · 2026-10-15
- Cloudflare's Clef decision models land on Merge Gateway, median latency 39ms — shensi · 2026-10-15
- Mine Bittensor Subnet 92 by pointing Claude Code at one link — airesearch12 · 2026-10-15
- Future agents will keep refactoring memory, building RL environments, and distilling to old checkpoints — manosaie · 2026-10-15