# Team Dax

- **Event:** [Built with Opus 4.7: a Claude Code hackathon](https://cerebralvalley.ai/e/built-with-4-7-hackathon)
- **When:** Apr 21 at 12:00 PM – Apr 27 at 2:00 AM (EDT)
- **Where:** Online
- **Team:** [Vago Daniel](https://cerebralvalley.ai/u/Dax)
- **GitHub:** https://github.com/daxaur/causalist
- **Demo video:** https://drive.google.com/file/d/1GNNUeOzgzdrpS3kTTnYkUZ2iZ6P8wayJ/view?usp=sharing
- **Gallery:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/246

Causalist makes any codebase legible — to humans and to agents.                     
                                                                                                                                                                                                                                                                                                                                                                 Drop in a GitHub URL and four Opus 4.7 agents (Structure, Dependency, Semantic, Oracle) draw your repo as a live causal graph in 30–60 seconds. Files become nodes, imports and calls become edges, every node is classified by  what it actually does (UI, API, logic, data, infra). For a beginner staring at an unfamiliar repo, it replaces hours of clicking through folders with a single picture of how the thing fits together. For a senior dev, it's the  mental model you'd build over a week, ready in under a minute.                                                                                                                                                                                                                                                                                                                                                                                                    But the bigger unlock is for coding agents. Today every agent operates blind: open one file, edit it, run tests, watch them fail, grep, repeat — burning context on rediscovery. Causalist gives agents a causal substrate. Through11 CLI tools (blast_radius, find_path, get_neighbors, summarize_node…), an agent can ask "what does this change break?" before changing anything, follow real cause-and-effect chains across files, and load only the slice of the repo that actually matters — slashing token usage and stopping the trial-and-error loops that eat sessions alive. Tree-sitter AST verification marks every LLM-inferred edge as confirmed or unverified, so the graph the agent reasons over is auditable, not hallucinated.                                                                                                                                                                                                                                One layer, two audiences: humans get a map of their codebase, agents get a verification loop for cross-file reasoning. Both think more causally, both burn less context, both stop guessing.

---

Markdown version of https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/246. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt