Use Relic with an AI agent

An AI agent in your terminal, such as Claude Code, can search your vault, save things into it, and pull a file out of it. It does that through the relic command, which stays on your computer and never talks to the cloud. What you install is a short text file, called a skill, that tells the agent the command exists and when to use it.

Install the skill

1. Get the command working first
   Everything here runs through relic. If the command is not found, The relic command covers how to get to it on Windows, on a Mac and on Linux.
   terminalCopy

   relic status

2. Install the skill
   This writes the skill to ~/.claude/skills/relic/SKILL.md and prints Installed the relic skill to and the path it used. Add --dir and a folder if your agent keeps its skills somewhere else. To read it first, run relic skill on its own and it prints the whole thing.
   terminalCopy

   relic skill --install

3. Start a fresh session
   An agent reads its skills when a session starts, so start a new one. After that it knows Relic is there, and it knows the sorts of question it should reach for Relic to answer.
4. Ask for something you copied
   Ask in ordinary words. The agent picks the command, runs it, and reads the answer back to you. You do not type any of this yourself.

What the agent can do

The skill points the agent at the same commands you would type, so it can do the same things you can. In practice that is four kinds of job.

• Find something you copied. A command, a link, a booking reference, the text inside a screenshot. It searches, then reads the item back.
• Save something for you. It can add an item with a title and tags, keep it in your vault, and put it back on your clipboard later.
• Tidy up what is already there. Add or remove your own tags, fix a title, add a note, keep an item or let it go back to the stream.
• Get a file out. Write a screenshot or a document from your vault to a path, so it can use it in whatever it is doing.

Anything it adds is an ordinary item. It appears in the Relic window straight away, the app titles and tags it in the background, and it syncs to your other devices like anything else.

What it cannot do without you

It cannot delete anything. Removing an item, removing a tag everywhere, and purging both refuse with exit code 6 unless you have allowed it yourself, and the skill tells the agent not to turn that on for you. If you do want something gone, allow it for that one command, and Relic still keeps a local copy first. The rules are on the command's page.

Things to ask for

These are the kinds of sentence the skill is written to recognise. There is nothing to memorise.

• "Find the tracking number I copied last week."
• "Save this command to my vault as deploy-prod."
• "What was that connection string I copied on Tuesday?"
• "Export the screenshot of the error to ./bug.png."
• "Tag everything I saved from the migration as ops."
• "Is Relic installed on this machine?"

For an agent that reads the web

The same skill text is served from relic.space/docs/cli-skill.md, so an agent that can fetch a page can read it without the command installed. A short index of the whole site, written for agents, is at relic.space/llms.txt. Every page here, this one included, is also available as plain markdown: add .md to its address.

There is no MCP server. The command and its skill are the supported way for an agent to work with a vault. The developer docs have the rest, including the sync API and how to run your own server.