Knowledge management · Agent Skill

Second-brain watcher, ready to edit and install

An agent watches everything you capture and delivers what bears on the questions you are still carrying, months after you asked them. You get the skill that runs it: read one real run, change it to fit your library, and install it on your own agents.

/second-brain-watch

1 What this does

You already wrote it down. That was never the problem.

A library is filed at write time and read at recall time, and the two rarely meet. You tag a note the way it made sense the day you wrote it. Months later the question you need it for is phrased differently, so you never go looking. Backlinks and folders connect what you already knew was connected, which is the part you did not need help with.

A watcher agent holds your open questions. Each question you are still carrying becomes a standing subscription on your own namespace. When something you capture bears on one, in none of the same words, it is delivered, together with the note where you first asked. You do not have to remember to go looking, because the library reaches you.

The same skill runs on a domain instead of a person. Point it at a team's namespace, where the material is design docs, tickets, incident write-ups, and decision records, and the open questions belong to the team. Nothing in the skill changes but the anchors.

Built from Semantik primitives

publish

Everything you capture, notes, highlights, meeting transcripts, voice memos, lands on your own namespace.

Each question you are still carrying becomes a standing SemQL subscription on that namespace.

On a match, the agent pulls what you already wrote, so the answer arrives with its history.

thread

Each question is a thread that accumulates as the library grows.

2 How the agent solves it

One real run, start to finish.

The same steps your own agents follow once you install the skill: the actual reasoning, including the messages it chose to ignore. Times count from the moment the first finding lands on the namespace.

Agent run · replayed from a real case

When

What the agent does

Setup

Your open questions become subscriptions

The agent reads the questions you have left open and stands one subscription per question. The anchor describes how an answer would read, in plain words, not how you phrased the question, because you asked it before you knew what the answer looked like. The contrast anchor is your own routine capture, so logistics never reach a question thread.

DISTANCE < 0.32 FROM "pairing new people with someone outside their own team is what made it last" CONTRAST "calendar change, room booking, meeting moved"

Seven months later

A line in a transcript lands

A transcript from an unrelated hiring review is captured to the namespace. One line in it reads "the teams that made it stick paired every new hire with someone outside their own function." It never uses the word onboarding. You had stopped looking in March.

Matched. The line embeds close to the answer anchor and far from the logistics contrast, delivered in the same moment it is captured.

Same minute

The answer arrives with its history

One line without its context is a fortune cookie. The agent searches the library and finds three things you already had: the note where you first asked, a book highlight from last spring, and a conclusion you half-formed and abandoned. It appends the new line to the question's thread with all three attached and a link back to the transcript.

An hour later

Routine capture is held back

You capture "moved the Thursday sync to 14:00, room 4 is booked." It lands on the namespace like everything else.

Not delivered. It embeds close to the logistics contrast and far from every open question, so your threads stay readable, no attention spent triaging it.

When you come back

The question is answerable now

You open the thread and read one question, four sources, and the date each one arrived. The connecting happened while you were doing something else. You write the conclusion yourself, close the question, and its subscription retires with it.

3 Make it yours

Start from the working version. Change what you need.

The skill below is the exact one the run above used. Pick where it runs, then edit it directly. Everything starts filled in, so you are always changing something that already works.

Platform

Namespace

Switching the namespace rewrites it inside the skill and the install commands.

Embedding model

—

Pinned by the namespace — — dimensions. Every publish and subscribe here uses it.

Skill · editable

0 lines

4 Connect it to your agents

Your agents get this skill the moment you connect it.

Install the Semantik MCP server once. Your skill rides on top of it. Pick the path that matches how you work.

From the terminal

For Claude Code or any MCP client

npx @noetive/mcp-server init --client claude-code --namespace global

npx @noetive/mcp-server doctor

One command does it: init writes the server entry and pins the namespace, and doctor says whether it took. Then drop your edited skill into the agent's skill directory. It calls the tools the server exposes: noetive_subscribe, noetive_publish, noetive_search.

Without the terminal

For Claude Desktop and connector-based clients

1
In noetive.io, open the namespace and copy its connection link.
2
In your client, open Settings → Connectors → Add and paste the link.
3
When asked, allow the connection. Your agents can now publish and subscribe on the namespace.
4
Paste the skill from the editor above into a new skill, or attach it to your project.

No keys to handle yourself. The connector holds the token, and you can revoke it from the dashboard anytime.

Test run

Replay the run from section 2: one message that should route, one that should not.

Start the agent on a loop with a namespace and a finish condition.

/loop 20m /second-brain-watch watch the global namespace, stop when every message is routed

Publish a message that should match from a second session. The agent routes it to whoever it affects on its next turn.

publish to the global namespace: "the teams that made it stick paired every new hire with someone outside their own function"

3
Publish noise that should be held back. The subscription drops it and the agent stays silent.
```
publish to the global namespace: "moved the Thursday sync to 14:00, room 4 is booked"
```

Both as expected? The skill is verified. If the message never arrives, check both sessions appear under Active Clients.

This is one task, done.

The stencil got you a working skill. Editing it moves it to your version of the problem. Soon you'll start from a problem description instead, and Semantik will draft the skill for you.