---
title: "Context for Halon — FireFight"
description: "Halon works from your resource map, service catalog, handbook, confirmed memories, incident history and connected tools. Each source stays current on its own, and your team decides what Halon keeps."
canonical: https://firefight.app/product/context
---

# Context

**Halon starts from what your team knows**

Your infrastructure, your owners, your rules and every past incident, in one place Halon reads before it answers. Connections keep it current. Your team keeps it right.

[Sign in with Slack](https://app.firefight.app/)

## Read live, confirmed by people

Connections refresh what can be read. People confirm what was learned. Anything a fresh reading contradicts is flagged before Halon relies on it.

### Read

**Always current.** Your connections keep the map up to date as things change. Handbook pages you keep elsewhere stay in sync on their own.

- Refreshed on its own
- Live updates from most providers
- Synced handbook pages
- Read failures reported

### Confirm

**Your team decides what Halon remembers.** Confirm or correct each lesson Halon learns. A handbook edit or a link Halon suggests waits until a person accepts it.

- Confirm or correct in Slack
- Handbook edits accepted first
- Suggested links confirmed
- Who decided, on the record

### Correct

**Live data beats old notes.** A fresh reading outranks any memory or handbook page. A memory that no longer matches is set aside or flagged for your team to check.

- Live readings win
- Contradictions raised at once
- Outdated memories flagged
- Reminders for what waits

## Six sources, one picture

Every chat and investigation starts from them, and Halon queries them as it works. Each one links to its docs.

- **Resource map** (Read from connections). What runs where, and what depends on what. Read live from your clouds, hosts, databases and code hosts. See [Resource map](https://firefight.app/docs/catalog/map.md).
- **Catalog** (Kept by your team). Services, teams, environments and product features, with owner teams and Slack channels. Each service links to the resources it runs on. See [Catalog](https://firefight.app/docs/catalog/overview.md).
- **Handbook** (Approved by your team). How your team releases, what counts as production, freeze windows and who directs Halon. Pages your team writes or accepts. See [Handbook](https://firefight.app/docs/ai/handbook.md).
- **Memory** (Confirmed by your team). Lessons from past incidents, postmortems and chats. Your team confirms, corrects or rejects each one. See [Memory](https://firefight.app/docs/ai/memory.md).
- **Incident history** (Recorded by FireFight). Past incidents, their timelines, findings and postmortems, with whether the team marked each answer right. See [Incident history](https://firefight.app/docs/postmortems/overview.md).
- **Connected tools** (Read live). Logs, metrics, traces, deploys, CI and code. Every read a connected tool offers, through FireFight. See [Connected tools](https://firefight.app/docs/api/integrations.md).

## Context that changes the first ten minutes

### Starts where the service runs

An investigation opens on the resources behind the incident's services and what changed around them just before it began.

### Learns from the last time

Past incidents that fired the same alert arrive with the answer each one got. A confirmed answer is a worked example, a wrong one is a mistake to avoid.

### Finds incidents by meaning

Halon searches past incidents, findings and postmortems that read like the current one, whatever words they used.

### Cites the handbook page

Every chat and investigation starts from the handbook. An answer that follows a page links to it, down to the section.

### Knows who owns it

The catalog names each service's owner team and Slack channel. Alert routing invites that team, and Halon reads ownership from the same record.

### Memory checked against live results

A memory is a starting point, never proof. Live results outrank it, and an answer flags every unconfirmed memory it relies on.

## Documentation

- [The map](https://firefight.app/docs/catalog/map.md): What goes on it, how it is read, links and live updates.
- [The catalog](https://firefight.app/docs/catalog/overview.md): Teams, services and owners, and how they meet the map.
- [The handbook](https://firefight.app/docs/ai/handbook.md): Pages, freeze windows, synced docs and the edits Halon proposes.
- [Halon's memory](https://firefight.app/docs/ai/memory.md): Where memories come from, and how your team decides on them.
- [Investigations](https://firefight.app/docs/ai/investigations.md): What a run starts from, including past incidents with the same alert.
- [Postmortems](https://firefight.app/docs/postmortems/overview.md): The record a completed incident leaves, and what Halon learns from it.
- [Connect an integration](https://firefight.app/docs/api/integrations.md): The tools Halon reads from, and what each is allowed to do.
- [Reading your platforms](https://firefight.app/docs/ai/reading-your-platforms.md): Every read a connected platform's API offers, with no approval needed.

## Give Halon the full picture

Connect your cloud and your code, add your services, and Halon starts from all of it. [Sign in with Slack](https://app.firefight.app/).
