GitHub Repository

Local-first memory for you and your coding agents. Every session and PR, searchable; the raw data is yours to analyze.

1 starsRust

Distributed LLM tracing and GH PR/issue linking [Apache 2.0]

Name: Distributed LLM tracing and GH PR/issue linking [Apache 2.0]
Availability: InStock
Author: supo

by supo·Jun 30, 2026·1 point·0 comments

Visit Project View on HN

AI Analysis

●●SolidSolve My ProblemCozy

Local-first agent memory with MCP server when Cursor and LangSmith already exist.

Strengths

•Local-first with no API keys—raw JSONL and SQLite index stay entirely on your machine.
•MCP server integration lets coding agents query session history over stdio.
•TUI built with charmbracelet/vhs for browsing sessions, PRs, and emergent topics.

Weaknesses

•Agent observability is getting crowded—Cursor, LangSmith, and others already track sessions.
•Core concept is straightforward recording and searching without novel technical approach.

Post Description

See where coding agents struggle, what tools eat tokens and how it all links to GitHub PRs & issues.

I open-sourced a tool that streams data from all your dev machines, VMs & sandboxes into S3. You can run your own analysis from there or use the bundled TUI.

Why I built this: I run a team of ~10 AGI-pilled engineers and struggle to keep track of how the work we do aligns to the milestones/projects we look in planning.

synty helps me identify areas of work that are getting a lot of agent/human attention but they are not accounted for in planning and we use this data to justify re-factoring/documentation efforts to remove friction for agents