# What is ContextOps?

> ContextOps is the operating discipline for the context layer AI agents use before they act.

ContextOps covers source selection, context capture, permissions, retrieval, packaging, freshness review, evaluation, and repair.

## What it includes

ContextOps includes source inventory, memory capture, access policy, redaction, retrieval, context packaging, freshness checks, evaluation cases, audit, and incident-style repair when the wrong context reaches an agent.

This is why a vector database alone is not ContextOps. Storage and retrieval are components; the operating loop is the product.

## How Ambience answers it

Ambience turns useful work from calls, tickets, PRs, docs, threads, and agent sessions into governed company context.

The same context then flows through approved agent runtimes through hooks, MCP, CLI, and the app, with source, scope, MemoryOps health, and audit attached.

## Related

- [ContextOps](https://ambience.sh/company-context/contextops)
- [ContextOps glossary](https://ambience.sh/glossary/contextops)
- [Context management for AI agents](https://ambience.sh/company-context/context-management-for-ai-agents)
- [Ambience vs internal RAG](https://ambience.sh/compare/internal-rag)