Install
$ agentstack add skill-kaisdavis-claude-skills-klaxon ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
klaxon: "I REALLY need your attention"
Triple-ping the macOS Submarine system sound with 1.333s pauses. Every fire is logged with timestamp, parent process, and reason to ~/.claude/logs/audio-hook.log, so the user can audit overuse.
Why this skill exists
Claude Code's default audio is silent. Many users wire a Stop-hook chime (every turn ends), and some wire a Notification-hook chime (when Claude Code is waiting on you). Both fire often enough that they fade into ambient noise. There needs to be a third tier: a sound the user reaches for once an hour at most, that genuinely cuts through.
That third tier is this skill. Triple Submarine is intentionally heavier than a normal alert. If you fire it for a non-emergency the user will tell you, and you should recalibrate.
When to fire (autonomous)
ALL three must be true:
- Forward progress is blocked on the user specifically. Not "I finished a turn." A real human-in-the-loop gate.
- Consequences exist if they do not see it soon. Destructive op queued, incident in flight, deploy gate, security finding, irreversible action pending.
- Tink (Stop) and Sosumi (Notification) would be too quiet for the gravity. If the situation would be served by a normal chime, use the normal chime.
Concrete examples:
- About to push to a production branch and need explicit ship approval. Fire BEFORE asking.
- Discovered live data loss or a running prod incident mid-task.
- A long-running operation has been blocked waiting on user input for more than two turns.
- About to take an irreversible action (force-push, db drop, mass delete) and need real-eyes confirmation.
Skip for:
- Every turn end. That is the Stop chime, automatic.
- "Done with the task." Silence is fine; the diff is the report.
- Routine clarifying questions. Those are Notification territory.
- Headless
claude -pcontexts. No human is watching, this is a no-op.
How to fire
~/.claude/skills/klaxon/scripts/klaxon.sh "short reason for the audit log"
The reason string lands in ~/.claude/logs/audio-hook.log under the KLAXON triple-ping header so the user can audit calibration. If you fired it for something that did not deserve three Submarines, expect to be called out and dial back.
Manual trigger
User typing /klaxon should fire ~/.claude/skills/klaxon/scripts/klaxon.sh "manual test". Smoke test for the chain.
Companion hooks (recommended bundle)
This skill is most useful as the loudest tier of a three-sound ladder:
| Tier | Sound | When | |---|---|---| | Quiet | Tink (1x) | Every assistant turn ends. Stop hook. | | Medium | Sosumi (1x) | Claude Code is waiting on user input. Notification hook. | | Loud | Submarine (3x, 1.333s pauses) | This skill. |
The companion script at scripts/play.sh (logger + headless suppression) plus the settings-recipe.json hook block wire up the quiet and medium tiers. See README.md in this skill's repo for the full install.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: kaisdavis
- Source: kaisdavis/claude-skills
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.