---
title: "Log what you learn"
description: "Use when you want your agent to keep what a project teaches in a manuscript. Covers creating a manuscript, how entries get written and how to see which manuscripts are active."
url: https://docs.heytrove.ai/usage/log-what-you-learn
updated: 2026-10-06
---

# Log what you learn

> Load this page when: you need to record project learnings in a manuscript

## For agents

- At the end of a task, compare what you learned with each manuscript's "log when" line.
- On a match, run `trove ms show <id>` and follow its instructions.
- If auto-update is on, write the entry and report it in one line. If it is off, show the entry and wait for a yes.
- If no manuscript matches, stay silent, or offer `trove ms create` when the topic will come back.

Keep every decision, pitfall and rule that you learn in a [manuscript](https://docs.heytrove.ai/concepts/manuscripts.md), so the next session can read it.

## Before you start

- Trove is installed (see [Install](https://docs.heytrove.ai/install.md)).
- You know what kind of learning the manuscript will hold, for example "a billing decision and its reason".
- You know if the agent may write entries without asking. You set this for each manuscript. See [Manuscripts](https://docs.heytrove.ai/concepts/manuscripts.md).

## Steps

1. **List the active manuscripts.** Ask your agent: "Show me my active Trove manuscripts." Check if one already fits your topic.
2. **Create a manuscript if none fits.** Ask your agent to create one. Give it a title and a "log when" line. The "log when" line says what kind of learning belongs in the book. You can also say how the agent writes an entry.
3. **Choose the approval rule.** Tell the agent if it may write entries without asking. If you want to approve each entry, say so. Many people keep approval on while the book is in progress.
4. **Work as usual.** At the end of a task, the agent compares what it learned with the "log when" line of each active manuscript. On a match, it writes or proposes an entry.
5. **Change the rule later.** Ask your agent: "Let the agent write Billing decisions entries without asking me." or "Ask me before each Billing decisions entry." Each manuscript has its own rule.

## Prompt to paste

```text
Create a manuscript called "Billing decisions". Log when: a billing decision and its reason. Instructions: one dated entry per decision, with the current state first. Do not write entries without asking me.
```

## What you will see

The agent creates the manuscript and prints it with its id and its book. Later, at the end of a task, the agent either reports one line about the entry it wrote or shows you the entry and waits for a yes. When the learning matches no manuscript, the agent says nothing.

```terminal
✓ Created manuscript: Billing decisions (ID: <id>)
  Book: <book-id>
```

## Reading your log later

When you need an old decision, ask your agent explicitly: "Search my Trove library for our billing decisions." That starts the [librarian](https://docs.heytrove.ai/subagents/librarian.md), which finds the manuscript book and reads the right pages. We are actively working on making the librarian start automatically and reliably, but today you should not rely on it.

## Direct CLI

Your agent runs these through the `trove` command. You can run them yourself in a terminal, but asking your agent is the recommended way: it picks the right flags, waits for processing and checks the result.

| Task | Command |
| --- | --- |
| List active manuscripts | `trove ms list --active` |
| Create a manuscript | `trove ms create` with a title, `--criteria`, and optional `--instructions` and `--auto-update` |
| Show a manuscript | `trove ms show <id>` |
| Change the approval rule | `trove ms update <id> --auto-update` or `--no-auto-update` |

See the [CLI reference](https://docs.heytrove.ai/reference/cli.md).

## Next

- [Manuscripts](https://docs.heytrove.ai/concepts/manuscripts.md)
- [Manuscripts that stay useful](https://docs.heytrove.ai/best-practices/manuscripts-that-stay-useful.md)
- [trove ms](https://docs.heytrove.ai/reference/cli/ms.md)