AI context is a set of notes on a story or bug where AI tools record what they've learned and done while working on it: the decisions behind the story, what was investigated, what changed in the code, and what needs testing. The next agent or person who picks up the item can read those notes instead of starting from scratch or digging through old chat sessions.

Most AI context is written by AI tools, either coding assistants and QA tools connected through the Atono MCP server, or the Story assistant in Atono. You can also read and edit it yourself in the web app. Together with the Glossary, which gives AI tools shared context across your workspace, AI context makes up Atono's product knowledge.

Add AI context

You can add AI context to stories and bugs, but not to drafts or epics.

On an item that doesn't have any AI context yet, the first content has to come from an AI tool. That's because the 'AI context' button only appears in an item's details panel once something has been added. After that, you can also add content manually to any empty tab.

Add AI context through the MCP server

📘
To use AI context from an AI tool, the tool must be connected to the Atono MCP server.

Any AI tool connected to the Atono MCP server can read and update AI context, including coding assistants in your IDE and QA tools that plan and run tests. When your assistant retrieves a story or bug, the MCP server instructs it to read the item's AI context before it starts work.

To keep AI context useful for the next session, ask your assistant to update it as you work, and include the item's Item ID (for example, STORY-98 or BUG-123) so it knows which item to update. When your assistant creates a story, or changes a story's user story or acceptance criteria in a way that shifts its scope, it may also record design decisions on its own, but mentioning it in your request is the most reliable way to make sure it happens.

In prompts, refer to each tab by the name the MCP server uses:

DESIGN for Design decisions
INVESTIGATION for Technical investigation
SUMMARY for Technical changes
QA_ANALYSIS for QA analysis

The following prompts are examples for common points in a session.

Keep AI context updated: At the start of a session, ask your assistant to maintain AI context as it works. For example:

“While you work on STORY-98, keep the AI context up to date. Use DESIGN for product and scope decisions, INVESTIGATION for research and findings, SUMMARY for implementation changes, and QA_ANALYSIS for QA findings, risks, and test coverage notes. Update AI context after each meaningful step.”

Capture design decisions: On stories, when making or revising product or scope decisions. For example:

“Update AI context for STORY-98. In DESIGN, capture the tradeoffs we discussed and why we chose this approach.”

Capture investigation notes: When discovering constraints, dependencies, or insights. For example:

“Update AI context for STORY-98. Add the key findings and constraints to INVESTIGATION.”

Capture implementation changes: After meaningful implementation steps, such as new behavior, refactors, tests, or schema changes. For example:

“Update AI context for STORY-98. In SUMMARY, record the changes you just made, including behavior and tests.”

Capture QA analysis: When planning what to test, or when a conversation or test run turns up an edge case worth covering. For example:

“Update AI context for BUG-123. In QA_ANALYSIS, note that the list needs to be tested with 10 or more entries, since that's when it crashed.”

Wrap up with a clean handoff: Before finishing work, so the next developer or agent can pick up without guessing. For example:

“Before we finish work on STORY-98, update AI context. Clean up DESIGN to reflect final scope, ensure INVESTIGATION lists remaining questions, confirm SUMMARY reflects the current implementation, and make sure QA_ANALYSIS lists what still needs testing.”

Add AI context with the Story assistant

When you're working on a story, the Story assistant can save decisions worth keeping to Design decisions. It does this when you accept suggested changes in the chat or accept a suggestion from a story review, and you can also ask it to save a specific decision. The Story assistant doesn't write to the other tabs.

Add AI context manually

Once an item has some AI context, you can fill in any of its empty tabs yourself.

Open a story or bug.
In the details panel, click AI context.
Select an empty tab.
Click Edit manually.
Enter the content, then click Save.

View AI context

Once a story or bug has AI context, you can open it from the 'AI context' button in the item's details panel. The button doesn't appear when you're viewing a bug in Bug triage.

Open a story or bug.
In the details panel, click AI context.
In the 'AI context' dialog, select a tab.

AI context is organized into tabs, one for each kind of content. Stories and bugs both have Technical investigation, Technical changes, and QA analysis. Stories also have Design decisions, which records the decisions made while writing the story.

A tab with content starts with a short note about what the tab is for, followed by the content itself, with the tab's versions listed on the right. Content is shown as plain text, so Markdown formatting such as headings and lists appears exactly as typed. An empty tab explains what belongs there and includes an 'Edit manually' button.

Design decisions

On stories, Design decisions captures the product and scope decisions made while writing the story with AI. It helps coding agents understand why a story is structured the way it is, and supports ongoing refinement as the story evolves.

Use it for:

Tradeoffs considered
Acceptance criteria framing decisions
Scope boundaries and constraints
Items explicitly deferred

This content should reflect current product intent, so update it as decisions evolve rather than treating it as a running log.

Technical investigation

Technical investigation captures discoveries, research findings, and constraints gathered while implementing the item with AI. It helps future agents avoid redundant work and build on what's already known.

Use it for:

Findings and constraints
Relevant files, endpoints, or dependencies
Open questions and decisions
Recommended next steps

When a finding is superseded or turns out to be wrong, it should be removed so the tab reflects what's currently known.

Technical changes

Technical changes records what AI tools have changed in the code for the item, as it currently stands. When a change is reverted, it should be removed from this tab rather than left in.

Use it for:

What was added, changed, or removed
Test updates
Behavior changes
Outcomes (what works now)

QA analysis

QA analysis records the thinking behind testing a backlog item: what should be tested, potential risks, and edge cases to cover. It lets QA tools and coding agents share that thinking, so the next agent can build on existing test planning when generating test plans and automated tests rather than starting from scratch. Product and scope decisions belong in Design decisions, not QA analysis.

Use it for:

What should be tested and why
Potential risks and edge cases to cover
Conditions that earlier test runs showed need testing
Test coverage notes, including known gaps

Like the other tabs, QA analysis should describe the current state rather than build up as a log. When an item's scope, acceptance criteria, or reproduction steps change significantly, the affected parts should be rewritten, while smaller changes only need the existing analysis updated.

Manage AI context

Anyone who can view a story or bug can edit its AI context and restore or delete its versions.

Edit AI context

Open a story or bug.
In the details panel, click AI context.
Select the tab you want to edit.
Edit the content, then click Save.

Saving creates a new version, which becomes the tab's current version. If you switch tabs or close the dialog before saving, Atono asks whether you want to keep editing or discard your changes.

Version history

Use version history to see how a tab's content has changed over time, or to recover content that a later update replaced.

Each tab keeps its own version history, listed on the right with the newest version at the top, labeled 'Current version'. Each version shows who saved it and when, and versions created by AI are marked with the sparkle (✦) icon. Its tooltip shows where the version came from, for example, 'Performed by MCP server' or 'Performed by Story assistant'. Select a version to view its content.

AI context has its own version history, separate from the item's content version history, and changes to AI context aren't recorded in the item's Activities.

Restore a version

Open a story or bug.
In the details panel, click AI context.
Select a tab.
In the version list, select the version you want to restore.
Click the ellipsis (...) icon and select Restore this version.

Restoring doesn't remove any history. It adds a new current version with the older content, and the version you restored from stays in the list. The current version doesn't have a restore option, since it's already in use.

Delete a version

Open a story or bug.
In the details panel, click AI context.
Select a tab.
In the version list, select the version you want to delete.
Click the ellipsis (...) icon and select Delete version.
In the confirmation dialog, click Delete version to confirm, or click X to cancel.

Deleting a version can't be undone, and it only affects the selected tab.

If you delete the current version, the next most recent version becomes current. If you delete the only version in a tab, the tab goes back to being empty, and if that leaves no content in any tab, the 'AI context' button no longer appears on the item.