/status

<objective>

Display the complete CAT hierarchy status with a visual tree showing all major versions, minor versions, and tasks with their current status.

Provides situational awareness for project progress.

</objective> <process> <step name="render-status">

Run the status script and capture output to a temp file:

"${CLAUDE_PLUGIN_ROOT}/scripts/status.sh" .claude/cat > /tmp/cat-status-output.txt 2>&1

This script handles BOTH data collection AND rendering with correct Unicode display widths.

If the script fails with "No planning structure found", inform user to run /cat:init.

Then use the Read tool to read /tmp/cat-status-output.txt and output the EXACT contents VERBATIM.

IMPORTANT:

• Do NOT rely on the Bash tool result display - Claude Code collapses large outputs.
• Do NOT summarize or paraphrase the file contents - output them EXACTLY as shown in the Read result.
• The user expects to see the complete box drawing with all lines, not a summary.

MANDATORY OUTPUT TEMPLATE (M145 - enforced after 2nd recurrence):

Your response MUST follow this EXACT structure:

[Read tool result shows file contents]

Your next message MUST be:

╭────────────────────────────────────────────────────────────────────────╮
│ [COPY ENTIRE BOX FROM READ RESULT - ALL LINES INCLUDING THIS ONE]     │
╰────────────────────────────────────────────────────────────────────────╯

---

## 🚀 NEXT STEPS
[then continue with next-steps section]

VALIDATION: Before outputting NEXT STEPS, verify your response contains the ╭ and ╰ box characters from the Read result. If it doesn't, you skipped the verbatim output step.

</step> <step name="next-steps">

After the status display, show the NEXT STEPS table:

🚀 NEXT STEPS

Option	Action	Command
[1]	Execute a task	/cat:work {version}-<task-name>
[2]	Add new task	/cat:add

---

Legend: ☑️ Completed · 🔄 In Progress · 🔳 Pending · 🚫 Blocked · 🚧 Gate Waiting

</step> </process>

<output_format>

The status output should be:

1. Generated by the status.sh script (NOT by LLM rendering)
2. Followed by NEXT STEPS table
3. Followed by Legend

</output_format>

<success_criteria>

• Script executed successfully
• Box display shown (from script output)
• NEXT STEPS table displayed
• Legend displayed

</success_criteria>

<history>

This skill was simplified in M142 (2026-01-19) to fix the 5th recurrence of status alignment issues (M136→M137→M140→M141→M142).

Root cause: LLMs cannot reliably calculate character-level padding for Unicode text.

Solution: Move ALL rendering to a bash script that uses Python for width calculation. The skill now just runs the script and outputs the result.

Previous versions had complex inline padding calculations, PRE-OUTPUT CHECKLISTs, and validation steps - all of which failed because they still relied on LLM padding calculations.

M143 update (2026-01-19): Changed to write script output to temp file, then use Read tool. Claude Code collapses large Bash tool results, so output must be shown via Read tool instead.

M144 update (2026-01-19): Clarified that file contents must be output VERBATIM, not summarized. "Display its contents" was ambiguous - agent summarized instead of echoing exact output.

M149 update (2026-01-19): Added explicit anti-pattern - NEVER manually type box characters. Even in other contexts (like /cat:work), agents must use scripts, not manual box typing.

M145 update (2026-01-19): Added MANDATORY OUTPUT TEMPLATE with validation checkpoint. M144's "VERBATIM" instruction failed - agent still skipped output. Added structural template showing exact expected format, plus validation check for box characters before NEXT STEPS.

</history>