Fair concern, but the base graph (graft build) is tree-sitter only, no LLM, free, works on any repo size. Summaries only run if you use --deep, and that's still experimental. graft init runs plain graft build by default, not --deep.
We're also still testing whether the summaries are worth it at all, whether the extra understanding they give graft ask actually cuts down on tokens, or if it's not worth the cost.","title":null,"type":"comment","url":null}],"created_at":"2026-08-06T15:56:44.000Z","created_at_i":1786031804,"id":49198429,"options":[],"parent_id":49197687,"points":null,"story_id":49197687,"text":"> Every source file is summarized once into a short description of what it does.
I admit I haven't had a chance to read the whole README, but wanted to get down my hesitation after I got pretty far (as an interested user): I think it'd either be too expensive or too-low-quality to feed all the source code in my repo into an LLM to build this dataset. Luna with medium+ thinking feels like my best bet right now but even then I'd worry about total token cost here for some repos.","title":null,"type":"comment","url":null}],"created_at":"2026-08-06T15:01:55.000Z","created_at_i":1786028515,"id":49197687,"options":[],"parent_id":null,"points":3,"story_id":49197687,"text":null,"title":"Show HN: Graft \u2013 Give coding agents a semantic map instead of grep","type":"story","url":"https://github.com/NanoNets/Graft"}