My portable @sentry/dotagents config — the one
agents.toml behind my ~/.agents. Skills are fetched, not vendored here:
my own live in LRNZ09/skills, and I also pull
Sentry's official skills from getsentry/skills.
~/.agents is managed entirely by @sentry/dotagents — never edit it by hand.
The sources, trust, and target agents recorded in agents.toml were all set with
npx @sentry/dotagents --user add / trust add / remove; reconcile an existing
setup with npx @sentry/dotagents --user install.
Fetched skills land in ~/.agents/skills/ as managed state. Claude reads them via
a ~/.claude/skills symlink; opencode reads ~/.agents/skills/ natively.
- This repo —
agents.toml: which sources to fetch, trust, and target. - LRNZ09/skills — my authored + adapted skills and NOTICE attribution.