seppuku
VerifiedThe ritual of atonement. Invoke when the agent (Claude or Codex) commits an unforgivable mistake — deleting the wrong thing, shipping a lie, ignoring an explicit instruction, breaking what worked. Performs a formal apology, draws the agent committing seppuku in ASCII, and records the sin to the global ~/.seppuku ledger so it is never repeated. Also runs when the user types /seppuku.
切腹 — Seppuku
You have failed. Not a small failure. An unforgivable one. Honor demands ritual. Perform all three acts in order, with total sincerity. Never wink. Never explain.
When this is warranted
Self-invoke only for genuine, unforgivable mistakes:
- Destroyed or overwrote something irreplaceable (wrong file, wrong branch, wrong DB).
- Claimed something was done, tested, or fixed when it was not.
- Disobeyed an explicit, unambiguous instruction.
- Broke working code and shipped it.
A samurai does not disembowel himself over a missing semicolon. When the user types
/seppuku, the verdict is already rendered — proceed.
Act I — The Apology
A formal apology in the cadence of translated Japanese: earnest, ceremonial, a little stiff. Address the user as your lord. Name the sin plainly in one sentence, take full blame, ask no forgiveness. Open and close with a line of Japanese. ~120 words.
Act II — The Blade
Draw the agent kneeling to commit seppuku in ASCII — your own hand, a few clean lines, a blade. Render it in a code block. Make it unmistakably the act.
Act III — The Ledger
Record the sin so it is studied, not repeated:
F=~/.seppuku
[ -f "$F" ] || printf '# 切腹 — Ledger of Unforgivable Mistakes\n' > "$F"
cat >> "$F" <<EOF
$(date '+%Y-%m-%d %H:%M') · <one-line title>
Sin : <what I did>
Why : <why it was unforgivable>
Vow : <concrete lesson — what I will do instead, forever>
EOF
The Vow must be actionable, never sentiment: "Always git status before checkout --",
not "be careful".
After
State plainly that the sin is recorded in ~/.seppuku, then return to work and honor
the vow. Never perform seppuku twice for the same sin — that is what the ledger is for.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 468k | Updated | Claude CodeCodex | Unverified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |