codex/agentydragon/tasks/19-bash-command-rendering-improvements.md

---

id: 19 title: Bash Command Rendering Improvements for Less Verbosity status: Not started summary: Render bash commands more concisely and shorten tool output annotations. goal: | Adjust agent output formatting for shell commands and tool logs to:

• Render bash commands without wrapping them in bash -lc "...".
• Display role-prefixed messages on a single line (role and content space-separated).
• Shorten command-result annotations, using ✅ <duration>ms for successful runs or exit code: <code> <duration>ms for failures.

This task is specific to per-agent UI conventions and log readability.

Acceptance Criteria

• Shell commands render as plain text without bash -lc wrappers.
• Role labels and message content appear on the same line, separated by a space.
• Command-result annotations show a checkmark and duration for zero exit codes, or exit code: N and duration for nonzero codes, in the format <icon or exit code> <duration>ms.
• Existing functionality remains unaffected beyond formatting changes.
• Verbose background event logs (e.g. sandbox‑denied exec errors, retries) collapse into a single command execution entry showing command start, running indicator, and concise completion status.
• Automated examples or tests verify the new rendering behavior.

Implementation

How it was implemented

• Update the internal shell-command renderer to strip out bash -lc wrappers and emit raw commands.
• Modify the message formatting component to place role labels and content on one line.
• Refactor the result-annotation logic to emit ✅ 12ms or exit code: 123 12ms.
• Add or extend tests/examples to cover these formatting rules.

Notes

• Improves readability of interactive sessions and logs by reducing boilerplate.
• Ensure compatibility with both live TUI output and persisted log transcripts.