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

# Turn stored information into useful context

> Context is the information your model receives for a specific task. Negentro helps your application assemble relevant information from persistent memory and configured sources instead of relying exclusively on the current prompt or replaying an entire conversation.

## Memory versus context

A user may have years of stored history, but a single request may require only a few preferences, facts, and recent decisions. Context assembly bridges that gap.

> **Memory**
>
> Information retained for potential future use. Can span years of user history across all sessions.

> **Context**
>
> The subset of information selected for the current operation. Task-specific, token-bounded, and ephemeral.

## How context assembly works

## Token-aware context

* Relevance to the current task.
* Source reliability and freshness.
* Token limits and reserved response capacity.
* Information priority and required instructions.
* Duplicates and conflicting facts.
* Privacy and authorization constraints.

## Context retrieval interface

```bash REST API theme={null}
POST /api/v1/context/retrieve
```

## Context quality checklist

## Recommended practice

Build context for the task at hand. Prefer a smaller, well-selected payload over a large collection of loosely related history. Keep retrieval and context assembly observable so you can tune quality as real workloads evolve.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.