> ## Documentation Index
> Fetch the complete documentation index at: https://docs.latcher.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Prompt Library

> Examples of effective prompts for Latcher.

A collection of reusable prompt patterns that help you get the most out of
Latcher’s multimodal research agent. Copy‑paste these into any Latcher chat,
tweak the details, and watch the system surface insights, audio, or visuals in
seconds.

### Kick‑start a Research Dossier

**When to use:** At the very start of an investigation. This prompt frames the
topic, scope, and desired outputs so Latcher can assemble a structured dossier.

**How to use:** State the topic, list the key questions, and specify which
Latcher outputs you want first (for example, an Insight Note or Context Map).

**Example Prompt**

```
Research Topic: regenerative medicine in gastroenterology
Key Questions:
- Latest clinical trials using stem‑cell patches
- Regulatory outlook in the US & EU
First output: create an **Insight Note** summarising the top three trials with
citations, then generate a **Context Map** linking the main institutions
involved.
```

### Generate a Concept Digest

**When to use:** Any time you need a dense, citation‑rich summary of a lengthy
paper, report, or multi‑file notebook.

**How to use:** Provide the source(s) and instruct Latcher to produce an Insight
Note at your preferred depth (overview vs. deep dive). Mention any blind spots
you want flagged.

**Example Prompt**

```
Please craft an **Insight Note** from:
- 2024 WHO report on antimicrobial resistance (PDF)
- My lecture notes (Notebook: “AMR Week 3”)
Depth: graduate level. Highlight conflicting stats between the report and my
notes, and flag any common misconceptions.
```

### Create an Audio Brief

**When to use:** Turning written material into a commuter‑friendly, two‑minute
dialogue.

**How to use:** Point to the notebook or files, set a target duration, and tell
Latcher which voice style to use (formal, conversational, etc.).

**Example Prompt**

```
Generate an **Audio Brief** from Notebook “Quantum Error Correction 101”.
Length: ≈2 minutes.
Tone: conversational but precise. Make sure each claim is cited, and end with a
one‑sentence takeaway.
```

### Visualise with a Context Map

**When to use:** When relationships matter more than linear text—debates,
literature reviews, concept hierarchies.

**How to use:** Give Latcher the source set and ask for a Context Map. You can
request clustering by theme, chronology, or influence.

**Example Prompt**

```
Create a **Context Map** of my “LLM Evaluation” folder.
Cluster nodes by evaluation dimension (helpfulness, safety, grounding) and
colour the clusters.
Export as PNG for my slide deck.
```

### Probe for Contradictions

**When to use:** To surface blind spots or alternative viewpoints before making
a decision or publishing a paper.

**How to use:** Ask Latcher to run the Contradictor agent on the current
notebook, then return a side‑by‑side list of assumptions vs. counter‑claims with
supporting sources.

**Example Prompt**

```
Run the **Contradictor** agent on Notebook “Carbon Capture Policy Memo”.
Output: table with columns → Assumption | Contradictory viewpoint | Best
supporting evidence (with links).
```

> **Tip**: These patterns are starting points—mix, match, and iterate to fit
> your workflow. If you craft a prompt that others could benefit from, share it
> in #resources on the community Discord so we can add it here!