supervision-audit

Classify user messages in Claude Code transcripts to identify wasted supervision and propose automation fixes.

Updated Nov 9, 2025
One-click install
npx skills add https://github.com/markusstrasser/skills --skill supervision-audit
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: supervision-audit
Source: https://github.com/markusstrasser/skills/tree/main/supervision-audit
Command: npx skills add https://github.com/markusstrasser/skills --skill supervision-audit

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Audit recent sessions for wasted supervision — corrections, boilerplate, rubber stamps — that should be automated. Outputs concrete fixes (hooks, rules, defaults). Run after a work day or after updating goals/constitution/CLAUDE.md.

Core Features & Use Cases

  • Deterministic extraction of supervision patterns from transcripts.
  • Synthesis and planning of fixes (hooks, rules, defaults, or new skills) with traceable outputs.
  • End-to-end improvement loop that integrates with CLAUDE.md and memory updates.

Quick Start

Run the audit after a work day to generate the latest supervision report and suggested automations.

Frequently Asked Questions about supervision-audit

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

FAQPage Schema
How do I automate supervision audits for Claude Code sessions?

You can automate supervision audits by running this Skill over recent Claude Code session transcripts. It deterministically extracts wasted supervision patterns and outputs concrete fixes like hooks, rules, or defaults to reduce manual toil.

What is wasted supervision in Claude Code transcripts?

Wasted supervision refers to repetitive corrections, boilerplate instructions, and rubber-stamp approvals in session transcripts that should be automated. The audit classifies these user messages into actionable automation opportunities to reduce manual toil.

How do I analyze Claude Code transcripts to generate automation hooks?

Run an end-of-day audit to perform deterministic pattern analysis on recent session transcripts. The audit synthesizes findings and proposes concrete fixes including automation hooks, rules, defaults, or new skills with traceable outputs.

Can I filter session transcript audits by specific projects?

Yes, you can apply optional project filters when running the supervision audit. This scopes the session analysis and automation opportunity classification to specific project contexts within your recent transcripts.

How do I update CLAUDE.md and memory files from session analysis?

Run the supervision audit to generate suggested automations and rules from session analysis. These traceable outputs integrate into an end-to-end improvement loop to update CLAUDE.md, CONSTITUTION.md, GOALS.md, or MEMORY.md.

When should I run a supervision audit on my session transcripts?

Run a supervision audit after a work day or after updating goals, constitution, or CLAUDE.md files. This timing ensures the session analysis captures the latest supervision patterns and generates relevant automation opportunities.