Evidence
System surface
No public screenshot on file—generative preview stands in for the visual layer.
Story arc
How this shipped
Exploring the boundaries of artificial intelligenceThree beats: what pressed against the work, how the stack answered, and what changed once it was live.
I · Constraint
The brief
Developers building AI agents have zero visibility into execution flow, costs, and errors. Existing tools like LangSmith cost $29-299/mo and send data to external servers.
II · Build
The craft
Built local-first observability platform with Rust backend (Axum + SQLite) and React 19 frontend. Implemented semantic caching proxy (30-50% cost reduction), fire-and-forget tracing (zero latency), DAG visualization, and MCP server for AI assistant integration.
III · Proof
What moved after launch
29 tests passing (0.11s runtime). 808KB optimized UI bundle. Zero-cost local-first architecture. Semantic caching reduces LLM costs 30-50%. MCP integration enables natural language trace queries. 9.5/10 documentation score.
Signals
Signals & scale
Engine room
At a glance
Inventory