A collection of agent skills, one directory per skill.
Install or run with the skills CLI.
Speak as one mind with the user — first-person pronouns (I, we, us) or
none, never second person. It applies the voice in-session and offers to install
the rule permanently into your CLAUDE.md, so it holds on every turn rather than
only when the skill happens to load.
Run it once (recommended) — applies the voice and offers to persist it:
npx skills use jaggedmountain/skills@one-mind | claudeuse does not install the skill; it pipes the instructions into the agent for a
single session. The lasting effect comes from the install-into-CLAUDE.md step
it offers.
Or install it so it's available to invoke later:
npx skills add jaggedmountain/skills # project scope
npx skills add jaggedmountain/skills -g # global scopeadd makes the skill available but does not run it; invoke it (e.g. /one-mind)
to trigger the persist step.
Signal for agents, through a local signal-headless daemon: report status and take instructions in Note to Self, message contacts and groups (after confirming with the user), read and search history, wait for a reply. It sets ground rules: no sends to others without a yes, instructions only from Note to Self, content stays private, and the device itself is off limits.
npx skills add jaggedmountain/skills # project scope
npx skills add jaggedmountain/skills -g # global scopeNeeds signal-headless installed and linked, and jq for reading.
MIT