What is this?
A set of agent skills for building with TypeSafe: they let your coding agent get typed decisions and probabilities from System One models.
A set of agent skills for building with TypeSafe: they let your coding agent get typed decisions and probabilities from System One models.Is this a library I import into my code, or something my AI agent loads?It's a pack of markdown skill files (SKILL.md) that coding agents load as a plugin — you don't import anything into your own code.
It's a pack of markdown skill files (SKILL.md) that coding agents load as a plugin — you don't import anything into your own code.Do I need Claude Code specifically, or do other coding agents work too?Any skills-compatible agent works: Claude Code gets a dedicated plugin, while other agents install through skills.sh, where you select your agent when prompted.
Any skills-compatible agent works: Claude Code gets a dedicated plugin, while other agents install through skills.sh, where you select your agent when prompted.How do I get my agent to actually use it once it's installed?Just ask in plain language, e.g. 'Use TypeSafe to route incoming support tickets by department, with human review for uncertain decisions.' In Claude Code you can also invoke the skill explicitly with /typesafe:typesafe-ai.
Just ask in plain language, e.g. 'Use TypeSafe to route incoming support tickets by department, with human review for uncertain decisions.' In Claude Code you can also invoke the skill explicitly with /typesafe:typesafe-ai.What does the typesafe-ai skill help with?It designs TypeSafe workflows, finds current docs and cookbooks, and helps you compose typed judgments in code.
It designs TypeSafe workflows, finds current docs and cookbooks, and helps you compose typed judgments in code.What does 'typed decisions and probabilities' mean in practice?System One models return decisions as typed values with confidence probabilities instead of free text — like routing a support ticket to a department and sending uncertain calls to human review.
System One models return decisions as typed values with confidence probabilities instead of free text — like routing a support ticket to a department and sending uncertain calls to human review.What are similar projects?evals-skills, claude-plugins-official, Intent-Router, headcount, pstack-claude, huashu-art-motion, ruflo, Bud-Decision-Studio
evals-skills, claude-plugins-official, Intent-Router, headcount, pstack-claude, huashu-art-motion, ruflo, Bud-Decision-StudioWhich topics is it in?ai-skills, system-one, claude-plugin, typesafe, typed-decisions, skills, coding-agent, plugin, markdown, agent-skills, claude-code, mcp
ai-skills, system-one, claude-plugin, typesafe, typed-decisions, skills, coding-agent, plugin, markdown, agent-skills, claude-code, mcp