You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Problem — TypeSafe's System One models (Jev) are now available via early access and are a candidate decision layer for amicode work (session-curation thread bucketing #1301, router-time curriculum switching #1295). The TypeSafe agent skill instructs installing its guidance into any project where it will be used. Approach — Vendor the upstream typesafe-ai/skills skill file verbatim into the project skills directory (.opencode/skills/typesafe-ai/SKILL.md), same pattern as the existing effect skill. Upstream source of truth: github.com/typesafe-ai/skills; re-sync on upstream updates. Scope — in: the skill file only · out: any SDK dependency, code, or integration (those follow via issue + PR if adopted).
Acceptance Criteria
.opencode/skills/typesafe-ai/SKILL.md matches the upstream raw file byte-for-byte.
Important
Problem — TypeSafe's System One models (Jev) are now available via early access and are a candidate decision layer for amicode work (session-curation thread bucketing #1301, router-time curriculum switching #1295). The TypeSafe agent skill instructs installing its guidance into any project where it will be used.
Approach — Vendor the upstream
typesafe-ai/skillsskill file verbatim into the project skills directory (.opencode/skills/typesafe-ai/SKILL.md), same pattern as the existingeffectskill. Upstream source of truth: github.com/typesafe-ai/skills; re-sync on upstream updates.Scope — in: the skill file only · out: any SDK dependency, code, or integration (those follow via issue + PR if adopted).
Acceptance Criteria
.opencode/skills/typesafe-ai/SKILL.mdmatches the upstream raw file byte-for-byte.