Claude Code 2.1.300
By Steven Van ·
Mods can display prompt-only text and switch agent views, while telemetry now separates cache writes by lifetime.
Mods gain more control over prompts and agent views in Claude Code 2.1.300. The release notes also add telemetry fields and advisor support on Claude Platform on AWS.
Prompt decorations and agent views
Mods can now use virtualText to draw text inside the prompt box without adding it to the draft or submitted prompt. A new prompt.change hook fires when the prompt text changes without a typed edit, including history recall, clearing, undoing or a plugin filling the box.
Calling $.ui.view({ agentId }) shows an agent's transcript in the main view; $.ui.view({}) returns to the previous view. The agent.spawn hook can also rewrite a workflow agent's prompt and model, where previously it could only refuse workflow agents.
A new dir property on a mod's Text supports right-to-left text. On Windows and in VS Code's terminal, its rows read in that direction and align flush right.
Cache and trace telemetry
The OpenTelemetry api_request event gains cache_creation_5m_tokens and cache_creation_1h_tokens, splitting cache writes by lifetime. Trace spans also gain GenAI semantic-convention attributes, including gen_ai.provider.name, gen_ai.conversation.id, gen_ai.tool.name and error.type.
Advisors on AWS
The /advisor command is now available on Claude Platform on AWS. No advisor runs until one is selected through /advisor, advisorModel or --advisor. Setting CLAUDE_CODE_DISABLE_ADVISOR_TOOL=1 disables the advisor tool.
