dos-stats

Aggregate DOS hook observations into structured activity reports with counts and latency metrics.

18|4|Updated Jun 10, 2026

Search Tags:#logging #audit #hooks #metrics #observability #latency

Ready to run

npx skills add https://github.com/anthony-chaudhary/dos-kernel --skill dos-stats

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: dos-stats
Source: https://github.com/anthony-chaudhary/dos-kernel/tree/main/claude-plugin/skills/dos-stats
Command: npx skills add https://github.com/anthony-chaudhary/dos-kernel --skill dos-stats

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill turns the DOS hook observation log into a readable summary so you can verify what the kernel actually did without manually counting events.

Core Features & Use Cases

Log folding: Converts .dos/metrics/observations.jsonl into counts, outcomes, exit codes, reason classes, stop blocks, delegates, and latency statistics.
Operational auditing: Helps confirm whether the native hook path is active, whether interventions are happening, and how often the system falls back to Python.
Read-only inspection: Useful when you need to review hook behavior over the whole workspace or narrow the view to recent activity without changing anything.

Quick Start

Ask the assistant to run dos-hook stats for the current workspace and summarize the resulting hook activity report.

Frequently Asked Questions about dos-stats

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I audit hook activity and latency metrics from a JSONL log?▼

Read-only auditing of hook activity aggregates the JSONL observation log into structured reports containing counts, reason classes, stop blocks, exit codes, and latency metrics. This verifies workspace hook behavior without manual event counting.

What is hook observability and how does log folding work?▼

Hook observability tracks system interventions by logging tool-call events. Log folding converts raw JSONL entries into aggregated summaries of outcomes, delegates, dialects, and latency statistics, providing a readable view of kernel actions.

Can I verify native hook paths and Python fallbacks without modifying the workspace?▼

Read-only inspection of the observations log verifies native hook paths and Python fallbacks without modifying the workspace. You can review entire workspace hook behavior or narrow the view to recent activity windows.

Do I need a specific binary and log file to generate hook metrics summaries?▼

Computing verb, outcome, exit-code, dialect, delegate, and latency summaries requires the bundled dos-hook binary and the local .dos/metrics/observations.jsonl log file to be present in the workspace environment.

What's the best way to summarize recent hook observations without manual counting?▼

The best way to summarize recent hook observations is using log folding to aggregate JSONL entries into activity reports. This automatically structures counts, outcomes, and latency metrics into a readable summary for quick verification.