Introduction to the site
This site is the first running output of an experiment: ship a personal
resume + daily log using open-weight LLMs, driven through local agent
tooling. The goal is not the novelty of AI code generation - it is building a
publishing loop where a machine drafts and I curate.
How it's made
Tech stack
Tooling
- OpenCode Go (opencode-go) for repo-aware agent runs.
- pi for orchestration / skills / MCP integration.
- Open-weight LLMs (local and API-routed) for drafting site code and
daily blog entries.
- Human (me) reviews, merges, and curates.
Why open-weight
- Inspectable prompts and outputs.
- Reproducible behavior; can pin a model version.
- No lock-in to a single vendor's content policy.
What ships daily
A cron-triggered agent writes 1-2 markdown posts from curated AI news,
commits them to the repo, and they appear here the next morning.