Chronicle

A local-first observability tool for AI agents, written in Rust. It records every LLM call and tool use, shows them on a timeline and as a graph, tracks token cost, and includes a caching proxy for OpenAI calls and an MCP server for querying traces.

Role: Creator
Years: 2025-2026
Status: Private repository
Stack: Rust
Axum
SQLite
React 19
TypeScript
Vite
Tailwind
Python SDK

Private repository

category: AI
stack: Rust, Axum, SQLite, React 19

No screenshot published2025-2026

The brief

When an agent run goes wrong, it is hard to see which call or tool use caused it, or what it cost.

The build

An Axum and SQLite backend, a React 19 front end, and a Python SDK. Traces stay on your machine.

What shipped

The repository is private and now archived.

Shares the most stack with this project

StanceStreamA real-time AI policy debate built for the Redis AI Challenge 2025. Agents debate over WebSockets, with Redis JSON, Streams, TimeSeries and Vector behind it, a semantic cache, and a fact-checking step.React 19
CodeCompassAn AI tool for learning an unfamiliar codebase: architecture diagrams generated from the code, and RAG-based Q&A over it. Archived.TypeScript
Finance QuestA financial literacy course with 18 chapters and more than 30 calculators. SM-2 spaced repetition schedules reviews, and an AI coach answers in context, using Claude with GPT-4o-mini as a fallback.React 19

Hiring for something like this?

Contract, advisory, or full-time. Tell me what you are building.

Start a conversation hello@elizabethannstein.com

The brief

The build

What shipped

Related work

Hiring for something like this?