Confirming Project Matches Tweet Request
Confirming Project Matches Tweet Request
Agent
Pi
Model
Claude Sonnet 4.6
StartedMar 28, 12:28 PM
Shared
Mar 28, 12:30 PM
Messages27
Downloads62
hellow clanker, can you get the details of this tweet: https://twitter.com/badlogicgames/status/2037811643774652911
The user wants to get details of a specific tweet. Let me try to fetch the tweet details. I can try using curl or other methods to get the tweet content. Let me try fetching the Twitter/X URL directly.
Actually, Twitter/X requires authentication to view most content now. Let me try a few approaches:
Let me try parallel_extract first since it's an external website URL the user wants to read.
Here are the details of the tweet:
Author: Mario Zechner (@badlogicgames) Posted: 8:38 AM · Mar 28, 2026
Tweet content:
"we as software engineers are becoming beholden to a handful of well funded corporations. while they are our 'friends' now, that may change due to incentives. i'm very uncomfortable with that. i believe we need to band together as a community and create a public, free to use repository of real-world (coding) agent sessions/traces. I want small labs, startups, and tinkerers to have access to the same data the big folks currently gobble up from all of us. So we, as a community, can do what e.g. Cursor does below, and take back a little bit of control again. Who's with me?"
He links to the Cursor blog post: Improving Composer through real-time RL
Engagement:
Quite a call to action — and very meta given this is the traces project! 👀
are we building here what's being ask of?
The user is asking if the traces project they're working in is what Mario Zechner (the author of that tweet) is asking for - a public, free to use repository of real-world coding agent sessions/traces.
Let me look at the project to understand what it does.
Let me take a quick look at the project to give you an informed answer.
>cat /Users/sergiydybskiy/src/traces/README.md 2>/dev/null | head -100>ls /Users/sergiydybskiy/src/traces/docs/>cat /Users/sergiydybskiy/src/traces/docs/adding-agents.md | head -60>cat /Users/sergiydybskiy/src/traces/docs/api.md | head -80Yes — you are literally building exactly what Mario is asking for.
Here's how it maps:
| What the tweet asks for | What traces is building |
|---|---|
| "public, free to use repository" | A platform for sharing, discovering, and discussing AI coding agent traces |
| "real-world coding agent sessions/traces" | CLI adapters that capture sessions from Claude Code, Cursor, Codex, Gemini CLI, pi, amp, opencode, etc. |
| "small labs, startups, tinkerers" | Self-hostable — open source, runs on Docker + Convex |
| "same data the big folks gobble up" | Shared @traces/shared types + REST API for programmatic access |
| "band together as a community" | Namespaces/workspaces, sharing, discovery — the social layer |
The irony is perfect: Mario (@badlogicgames) posted this call to action today, and you're already deep into building it. The share-to-traces skill even exists as a registered skill in this repo's agent instructions — so Pi itself can share its own sessions to Traces automatically.
You might want to reply to that tweet. 😄
thanks clanker! can you share this trace plz
The user wants to share the current coding session to Traces. I have a skill for that: share-to-traces. Let me read the skill file first to understand how to do it.
The skill says to use traces share --cwd "$PWD" --agent gemini-cli --json but this is a Pi session, not Gemini CLI. Let me check what agent I should use. Looking at the AGENTS.md, the agents include 'pi'. So I should use --agent pi instead.
Let me first check if traces CLI is available and what options it supports.
>traces share --list --cwd "$PWD" --agent pi --json 2>&1I can see the current session is the first one with id 179633ae-2c84-4e12-a70e-31df9e0b74dd - it starts with "hellow clanker, can you get the details of this tweet". Let me share that one.
Found the current session. Sharing it now!
>traces share --trace-id 179633ae-2c84-4e12-a70e-31df9e0b74dd --json 2>&1