Context and artifacts

Keep agent chats focused with lane context, attachments, proof, and session history.

ADE keeps chat context tied to the selected lane and session. The agent can inspect the repository directly when it needs more detail.

What the agent sees

• Your prompt.
• The selected lane and branch.
• Relevant PR or Linear context when linked.
• Attachments you add to the chat.
• Recent session context when you resume a conversation.

Attachments and artifacts

Use artifacts when they make the task clearer:

Artifact	Example
Screenshot	UI state the agent should match or debug.
Log	Error output from a failing command.
Proof	Evidence captured after a change.
PR comment	Review feedback the agent should address.
Linear issue	Requirements or acceptance criteria.

Context window

Long chats eventually fill the model context window. When that happens:

• Continue if the agent is still on track.
• Ask the agent to summarize the current state.
• Start a new session in the same lane with the summary.
• Attach the key logs or screenshots instead of relying on old messages.

Compaction and summaries are useful, but they are not exact memory. Restate any critical constraint before asking the agent to continue.

Find old context

Use History to reopen transcripts, inspect tool calls, and recover proof or PR links from previous sessions.

History

Search past sessions.

Proof

Capture and attach evidence.