skills.rest
Yassimba
YassimbaCOMMUNITY
Explore all skills by this author →
GitHub

skill-doctor

Scores local agent conversations against rubrics and drafts skill improvements with a shareable HTML report.

9|Updated Jul 17, 2026
Search Tags:#code quality#claude code#report generation#agent evaluation#skill grading#conversation scoring#efficiency rubric
One-click install
npx skills add https://github.com/Yassimba/loom --skill skill-doctor-yassimba
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: skill-doctor
Source: https://github.com/Yassimba/loom/tree/main/skills/skill-doctor
Command: npx skills add https://github.com/Yassimba/loom --skill skill-doctor-yassimba

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve? It is hard to know whether installed agent skills actually help. This Skill grades your agent setup from real local conversation history, scoring efficiency and code quality, then proposes concrete, evidence-backed skill edits. ## Core Features & Use Cases - Conversation Scoring: Samples recent Claude Code, Codex, Pi, and Warp sessions and scores them against efficiency and code-quality rubrics. - Skill Improvement Drafts: Writes proposed SKILL.md edits with unified diffs, each traced to a failed conversation as evidence. - Shareable Report: Renders a self-contained HTML report with a letter grade, score bars, findings, and a share-as-PNG button. - Use Case: After a month of coding with an agent, run the skill to learn that your debugging skill never triggers, and receive a rewritten trigger description plus a diff you can apply. ## Quick Start Ask the agent to use the skill-doctor skill to grade my recent agent conversations and suggest improvements to my installed skills.

Frequently Asked Questions about skill-doctor

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

FAQPage Schema
How do I grade my AI coding agent's performance?▼

Run the skill-doctor skill, which samples recent local agent conversations and scores them against efficiency and code-quality rubrics. It produces a letter grade, three top findings, and a shareable HTML report.

How to tell if my installed agent skills are actually working?▼

The skill measures skill coverage, the fraction of sampled sessions where an installed skill was detected, and flags skills that never triggered in failed conversations. It then drafts trigger-description fixes with diffs.

Which coding agents does skill-doctor support?▼

The collector reads session history from Claude Code, Codex, Pi, and Warp local storage. The --harness flag selects a source, and auto mode scans every locally available one.

Does skill-doctor upload my conversation transcripts?▼

No. Everything runs locally; transcripts and session files are never uploaded. The only shareable artifact is the report HTML or PNG the user explicitly chooses to export.

What happens when no recent conversations are found?▼

If the sampled session count is zero, the skill reports there is nothing to score in the selected scope and suggests raising the --days lookback window or choosing different projects, then stops.

Related Skills

Explore More Software Engineering Skills→
build000rbuild000r
5

prompt-reviewer

Score conversation prompts from Claude Code, Codex, AMP, and OpenCode using a 9-axis rubric.

Community
Advanced
mintuzmintuz
3

skill-doctor

Grades agent conversations against efficiency and code-quality rubrics and drafts skill improvements.

Community
Advanced
agent0aiagent0ai
18.8k

create-skill

Generate structured SKILL.md files with required frontmatter fields.

Official
Intermediate
StudyDriftStudyDrift
58

control-cli

Automates creation and reuse of deterministic local harnesses for interactive CLI/TUI sessions.

Official
Intermediate