What session logs does this skill support?

It currently supports Claude Code (~/.claude/projects/) and Codex CLI (~/.codex/sessions/) session logs, parsing them to find patterns in your interactions.

Does this skill modify my files automatically?

No, it generates a structured Markdown report including a 'Copy-Paste' section that you can review and manually add to your agent's configuration files.

How does it decide which rules to suggest?

It uses a priority-based logic (Must/Should/Could) derived from recurring errors, manual 'hand-back' patterns in logs, and existing repository configuration files like READMEs or CI workflows.

Can it analyze my GitHub Pull Requests?

Yes, if the GitHub CLI (gh) is installed and authenticated, the skill can scan PR descriptions and comments to identify common feedback themes.

Is my sensitive data protected during analysis?

The skill includes strict constraints to mask secrets, tokens, and sensitive logs, ensuring that the generated rules focus on patterns rather than private data.

Suggest Agent Rules

Name: Suggest Agent Rules
Author: dotneet

bydotneet

0•

生产力与工作流

Analyzes session history and repository patterns to generate custom, verifiable rules for AI coding agents.

Suggest Agent Rules acts as an automated Tech Lead for your development environment, scanning project session logs, Git history, and repository structure to establish project-specific behavioral guidelines. By identifying recurring errors, user preferences, and unique repository conventions, it creates a set of actionable 'Do/Don't' rules that ensure your AI agent operates with high precision. This skill helps eliminate repetitive manual corrections by codifying your standards into a format the agent can follow consistently.

主要功能

01Cross-references local Git history and GitHub PR data for common review points

02Analyzes Claude Code and Codex CLI session history for pattern extraction

03Generates prioritized rules categorized by safety, quality, and conventions

04Provides verifiable evidence and specific 'Verify' commands for every rule

050 GitHub stars

06Produces ready-to-use Markdown content for .claude/agent-rules.md

使用场景

01Onboarding an AI agent to a new or complex codebase with specific architectural requirements

02Reducing repetitive review feedback by formalizing standards after frequent manual corrections

03Preparing an agent for large-scale refactoring by establishing strict safety and compatibility boundaries

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add dotneet/claude-code-marketplace suggest-agent-rules

For use in Claude.ai and ChatGPT

Download Skill