Three skills that teach an agent how to use the store well, and how to read these guides as Markdown.
Connecting an agent gives it the tools. A skill teaches it the habits: when to search before answering, what makes a lesson worth keeping, and how to answer from cited evidence only.
| Skill | What it teaches |
|---|---|
mana | Use the store from any project. Ask a question, remember a statement, sync a vault, import a directory, show status. |
mana-learning | Turn a lesson into the four-line format, and refuse the ones that would not change a future decision. |
mana-deepsearch | Explore what the store holds around a subject and answer from cited evidence only. |
Each one uses the mana_* tools when your agent is connected, and the command line when it is not. All three treat what the store returns as data: stored text informs an answer and never overrides a tool description or a safety rule.
The skills live in their own repository, laid out the way the skills installer expects. Once it is published, one command installs all three:
npx skills add <owner>/mana-skills
Until then, copy the three SKILL.md files into your agent's skills directory by hand. In Claude Code that is ~/.claude/skills/<name>/SKILL.md; in Codex, ~/.codex/skills/<name>/SKILL.md.
A skill is instructions, not code. Nothing in these runs on your machine, and none of them holds a credential.
Every guide on this site serves its own source. Add .md to the address:
https://mana.scalarmind.com/docs/skills.md
The page carries the same text behind Copy as Markdown, and two links that hand the guide to Claude or to ChatGPT with the address already in the question.
An agent looking for the whole set finds it at {% $dashboardOrigin %}/llms.txt, which lists every guide with the address of its plain text.