# Inspect Scout

> Transcript Analysis for AI Agents

## Getting Started

- [Getting Started](https://meridianlabs-ai.github.io/inspect_scout/index.html.md): Transcript Analysis for AI Agents

## Using

- [Scanners](https://meridianlabs-ai.github.io/inspect_scout/scanners.html.md): Basics of using scanners
- [Transcripts](https://meridianlabs-ai.github.io/inspect_scout/transcripts.html.md): Reading and filtering transcripts for scanning
- [Workflow](https://meridianlabs-ai.github.io/inspect_scout/workflow.html.md): End to end workflow for transcript analysis projects
- [Projects](https://meridianlabs-ai.github.io/inspect_scout/projects.html.md): Using projects to manage scanning configuration
- [Results](https://meridianlabs-ai.github.io/inspect_scout/results.html.md): Collecting and analyzing scanner results
- [Validation](https://meridianlabs-ai.github.io/inspect_scout/validation.html.md): Validation of scanner results against ground truth target values

## Scanners

- [LLM Scanner](https://meridianlabs-ai.github.io/inspect_scout/llm_scanner.html.md): Customizable LLM scanner for model evaluation of transcripts
- [Grep Scanner](https://meridianlabs-ai.github.io/inspect_scout/grep_scanner.html.md): Customizable scanner for text-pattern searching of transcripts
- [Custom Scanners](https://meridianlabs-ai.github.io/inspect_scout/custom_scanner.html.md): Comprehensive documentation on creating custom scanners
- [Scanner Tools](https://meridianlabs-ai.github.io/inspect_scout/scanner_tools.html.md): Create custom scanners with the same tools used under the hood by `llm_scanner()`
- [Multi Agent](https://meridianlabs-ai.github.io/inspect_scout/multi_agent.html.md): Scanning transcripts with multiple agent trajectories

## Transcripts

- [Transcripts Database](https://meridianlabs-ai.github.io/inspect_scout/db_overview.html.md): Overview of transcript database concepts and basic usage
- [Database Schema](https://meridianlabs-ai.github.io/inspect_scout/db_schema.html.md): Details on the required and optional fields for databases and underlying data formats for messages and events
- [Capturing Transcripts](https://meridianlabs-ai.github.io/inspect_scout/db_capturing.html.md): The `observe` decorator and context manager captures LLM calls and writes transcripts directly to a database
- [Importing Transcripts](https://meridianlabs-ai.github.io/inspect_scout/db_importing.html.md): Import transcripts into databases from a variety of sources
- [Publishing Transcripts](https://meridianlabs-ai.github.io/inspect_scout/db_publishing.html.md): Best practices for publishing transcript databases for use by others