Second Brain Prompt: Capture, Retrieve, and Update Context

A second brain prompt helps you decide what is worth capturing, how to retrieve what you saved, and how to keep the record current as the work changes. The challenge is not storing more, it is storing the right things in a form that is useful when you come back to them. Tools like fenbs keep the resulting work record, including decisions, context, and where you left off, in one place that any connected AI can read and update.

The templates here cover three moments in a second-brain workflow: capturing after a session, retrieving before a new one, and updating a record that has gone stale.

What belongs in a second brain

A second brain that stores everything becomes a search problem. The record that is useful to a future AI session is specific: the conclusions you reached, the decisions that still hold, and the next step that makes sense given where you are. Source material is valuable when it is hard to recreate; summaries are valuable when they carry the reasoning, not just the result.

• Decisions: what was chosen, who chose it, and why it was chosen over the alternative.
• Open questions: what is still unresolved and who needs to answer it.
• Where we left off: the specific next step and what is still in progress.
• Lessons: what happened, what was learned, and what to do differently.
• Links to source material: where the important evidence or reference lives.

A prompt for capturing after a session

The best time to run a capture prompt is at the end of a work session, before the context is clear in your head but before you have forgotten the details. A good capture prompt asks you or your AI assistant to record the conclusions, not the process that led to them.

Help me record what happened in this session.

Project: [project name]
What we worked on: [brief description]

Record the following:
1. Decisions made (what was decided, by whom, and the reason)
2. Open questions (what is unresolved and who should answer it)
3. Where we left off (the specific next step and anything still in progress)
4. Lessons (anything that should change how we approach this next time)

Keep each entry short and specific. Do not summarize the process, record the conclusions.

For the broader question of what belongs in a second brain versus a note, what is an AI second brain explains the difference between storing knowledge and storing the work. For a longer exploration of the AI brain concept, what is an AI brain covers how context outside a chat session becomes usable.

A prompt for retrieving before a new session

A retrieval prompt runs at the start of a session to reconstruct the context a fresh assistant needs to continue the work. The goal is not a summary of everything, it is the minimum context required to take the right next step without asking the same questions again.

Read the project record for [project name] and give me a briefing to start the next session.

I need to know:
1. What was decided since the last time I worked on this
2. What is still open and waiting for a decision
3. Where we left off: the specific next step
4. Any lessons or rules that apply to the work I am about to do

Keep the briefing under 200 words. Flag anything that has changed since I last worked on this.

A prompt for updating a stale record

A record that is not updated quickly becomes something nobody trusts. When an AI session ends without a save, or when a decision changes and the old record is not updated, the next session can read stale context and proceed in the wrong direction. An update prompt runs when the record needs to catch up with where the project actually is.

Update the project record for [project name].

Here is what has changed since the last update:
[describe the changes: completed tasks, new decisions, changed direction, resolved questions]

For each change:
- Update the relevant decision or note to reflect the current state
- Close any open question that was answered
- Add new open questions that appeared
- Update Where we left off to reflect the current next step

Mark any part of the old record that is now out of date.

Keeping the record useful across AI sessions

The hardest part of a second brain is keeping it current enough to be trusted. A record that is accurate on Monday and not updated until Thursday has four days of decisions and changes buried in conversations. The capture prompt runs best as a habit at the end of each session rather than as a catch-up at the start of the next one.

When you change AI tools or start a new session in a different app, the record outside the chat is what carries the work forward. The context does not live in the chat; it lives in the record the chat produced. Portable AI memory covers how to make that record survive switching tools. AI memory for teams covers the shared version, where multiple people and assistants use the same record.

The smallest useful second brain record answers three questions: what was decided, what is still open, and what is the next concrete step. If a record answers all three clearly, a new session can start useful work immediately. If any of the three is missing, the session begins with reconstruction rather than the actual work. Write each capture prompt output as a direct answer to those three questions, and check before you close the session that each answer is specific enough to be actionable. Vague language at capture time becomes confusion at retrieval time. A decision recorded as "we agreed on the approach" is not usable; one recorded as "we chose option B because of cost; the next step is to draft the contract" is.

How to do this with fenbs

1. Create a free account and add a project for the work you want to track.
2. Run the capture prompt at the end of each session and write the decisions, open questions, and next step into the project.
3. Save the capture, retrieval, and update prompts in Skills and prompts for easy reuse.
4. Connect your AI assistant over MCP from Integrations so it reads the project record at the start of each session.
5. Run the update prompt whenever a decision changes or the project takes a new direction.

fenbs does not import notes from Notion, Obsidian, or other note tools. It keeps the work record: decisions, context, tasks, and where you left off, available to any connected assistant.