HackerTrans
トップ新着トレンドコメント過去質問紹介求人

syumpx

no profile record

投稿

Show HN: Sayou: Open-source agent workspace with versioned files and MCP tools

github.com
3 ポイント·投稿者 syumpx·5 か月前·1 コメント

Sayou – Open-source Dropbox for AI agents

6 ポイント·投稿者 syumpx·5 か月前·4 コメント

コメント

syumpx
·5 か月前·議論
Agents today either forget everything when the session ends, or remember fragments out of context through vector search. Neither is how real work gets done.

I built sayou because I was tired of my agents losing their work between sessions — research reports, project docs, client records, all gone. Vector DBs can retrieve similar content but you can't version, diff, or structure what's stored there. Traditional databases force you to define schemas upfront for work that's inherently messy.

sayou is a file-based workspace for AI agents. Files with YAML frontmatter for structure, markdown for content, append-only versioning, full-text search, and an audit trail. It connects to any MCP-compatible client (Claude Code, Cursor, Windsurf) with 11 tools out of the box.

Stack: Python, SQLite for local dev, MySQL + S3 for production. No external dependencies for the base install.

What it looks like in practice: an agent writes research to research/competitor-pricing.md with status: draft in frontmatter. Next session, another agent searches for it, reads it, updates the status to reviewed. Every write is a new version. Nothing is overwritten.

Quick start: pip install sayou && sayou init --claude

Interested in the broader question too — how are you handling persistent state for multi-session agents? We think "agent workspace" is a missing category between memory systems and databases.
syumpx
·5 か月前·議論
Current approach is last-write-wins with version history - simple but doesn’t solve concurrent edits.

I don’t think auto-merge is the right default for prose/research (unlike code where Git’s merge works). When Agent A writes strategic memo concluding X while Agent B writes concluding NOT-X, merging both is worse than surfacing the conflict.

Thinking the right model is: ∙ Optimistic writes (current behavior) for most cases ∙ Explicit locks for high-stakes docs agents know they’re collaborating on ∙ Diff tooling for post-hoc resolution when conflicts do occur

thanks for asking a great question. whats your thoughts?
syumpx
·5 か月前·議論
multi step may be whats killing it. simply and let llm do the work
syumpx
·5 か月前·議論
congrats!
syumpx
·5 か月前·議論
Thanks! We just shipped a Claude Code plugin so setup is now one command:

  claude plugin install sayou@pixell-global
                                                                                                                                 
It adds hooks that automatically show your workspace at session start and capture your work in the background. So that "it actually remembered" experience you got works without even asking it to save.