CodeSeeker – Knowledge graph code intelligence for AI coding assistants
Knowledge graph + semantic search for AI—but 'chat with your codebase' already saturated.
Unity-aware AI infrastructure for Claude Code — a knowledge graph + 88 MCP tools that let your AI agent know your project, not just grep its files.
88 MCP tools map Unity prefab inheritance that stock Claude Code misses.
Unity game developers using AI coding assistants
Cursor · Continue · Sourcegraph Cody
Straightforward grep-and-guess is very weak in terms of efficiency, and — most importantly — in terms of reliability. Claude misses the relations between entities in your game, that's the bitter truth. If you're interested in an actual comparison, concrete example: I asked the agent "which prefabs and scenes break if I change EnemyAI?" Stock Claude Code read ~200k tokens of YAML, found 1 prefab, missed 3 variants, and suggested a change that breaks inheritance. With Hades it returned the correct map — 4 prefabs, 3 scenes — in 7 tool calls, 27% cheaper. Full side-by-side with repro: https://github.com/TheArcForge/Hades/blob/main/Documentation...
Since I work with Unity and Claude daily as part of my job, I came up with a solution. Hades builds a graph out of your project's entities, and Claude can query it via MCP tools.
That's the simple version — if you want the details, I wrote up the full story on Medium: https://medium.com/@mike.kuharuk/unity-context-graph-for-cla...
So far I've only proof-tested it on macOS and Unity 6; that scope will grow step by step. Also v1 is mainly built around Claude Code, so Cursor and all other AI agents supports will be added post launch.
I'd appreciate technical feedback and suggestions.
Knowledge graph + semantic search for AI—but 'chat with your codebase' already saturated.
Postgres + pgvector + MCP integration solves the AI amnesia problem better than vector DB silos.
Portable AI memory across Claude, GPT, Gemini using your own Postgres knowledge graph.
Knowledge graph beats vector search for Claude Code context injection.
Dialectic thesis-antithesis-synthesis research loop is novel but clearly alpha software.
Consolidates scattered AI memory from Claude, ChatGPT, and MCP into one searchable space.