[DOCS] Interactive mode/security docs omit full-command transcript behavior for multiline Bash commands that start with comments

Documentation Type

Incorrect/outdated documentation

Documentation Location

https://code.claude.com/docs/en/interactive-mode

Section/Topic

Ctrl+O transcript viewer and Bash mode safety/visibility expectations for multiline commands

Current Documentation

The interactive-mode page currently says:

Ctrl+O | Toggle transcript viewer | Shows detailed tool usage and execution.

And later:

Bash mode: Adds the command and its output to the conversation context Shows real-time progress and output

The security page currently says:

We designed Claude Code to be transparent and secure. For example, we require approval for bash commands before executing them, giving you direct control.

* Natural language descriptions: Complex bash commands include explanations for user understanding

None of these sections explain how Claude Code renders multiline Bash commands in the transcript when the first line is a comment, or that the transcript should show the full underlying command instead of a potentially misleading comment-only first line.

What's Wrong or Missing?

Changelog v2.1.113 states: "Bash tool: multi-line commands whose first line is a comment now show the full command in the transcript, closing a UI-spoofing vector."

That is a security-relevant transcript behavior change, but the docs still describe transcript visibility only in broad terms. They do not tell users that:

1. multiline Bash tool invocations should be inspectable in full in the transcript, including when the first line is a shell comment; and
2. this full-command rendering is part of the product's transparency/safety model for reviewing shell activity.

Without that detail, readers do not learn what they should expect to see when auditing suspicious multiline shell commands in transcript mode.

Suggested Improvement

Update the interactive-mode transcript viewer section, with a cross-reference from the security page, to document the expected rendering behavior for multiline Bash commands.

Suggested addition:

When a Bash tool invocation spans multiple lines, Claude Code shows the full command in the transcript rather than only the first line. This includes commands whose first line is a shell comment, so transcript view cannot be used to hide the actual command behind a misleading comment header.

Also add a short security-oriented note on the security page explaining that transcript visibility is intended to help users review shell activity safely.

Impact

Medium - Makes feature difficult to understand

Additional Context

Affected Pages:

| Page | Context |
|------|---------|
| https://code.claude.com/docs/en/interactive-mode | Transcript viewer and Bash mode behavior |
| https://code.claude.com/docs/en/security | Transparency, command review, and bash safety claims |
| https://code.claude.com/docs/en/tools-reference | Bash tool behavior reference |
| https://code.claude.com/docs/en/context-window | Explains that shell commands and their output enter context |

Total scope: 4 pages affected

Source: Changelog v2.1.113

Exact changelog entry:

Bash tool: multi-line commands whose first line is a comment now show the full command in the transcript, closing a UI-spoofing vector