AI Agents Need Artifacts, Not Activity
I got tired of AI agents sounding busy and leaving nothing behind. Here's the boring rule I now use before calling agent work done.
Ephemeral Sharing encrypts files and text in your browser, stores file contents as ciphertext in R2, and supports expiring links with best-effort download limits.
AI agents need structure from config files, but they should not need raw secrets in their context. ReadSafe is my pre-alpha attempt to fix that locally.
I got tired of agents handing me giant Markdown blobs I would never read properly. So I built MaraDocs to turn agent reports into durable HTML artifacts.
AI agents aren't magic. They're a loop. Here's the full picture — with a live dashboard you control.
Spoiler: you probably don't need to. Here's what actually happens after you integrate an LLM API, from whispering to a dragon to raising one from an egg.