qiaomu-goal-meta-skill  by joeseesun

Refining vague agent tasks into executable goals

Created 1 week ago

New!

651 stars

Top 50.6% on SourcePulse

GitHubView on GitHub
Project Summary

This project addresses the common issue of AI agent task failures stemming from vague or poorly defined objectives. It targets developers and users interacting with AI agents like Codex, transforming ambiguous requests into precise, actionable /goal commands. The primary benefit is significantly increasing the reliability and success rate of AI-driven tasks by providing structured, executable instructions.

How It Works

The qiaomu-goal-meta-skill acts as a pre-processor for AI agent commands. It takes a user's high-level, often vague, requirement and systematically expands it into a detailed /goal command. This command includes explicit definitions for outcome, verification procedures, operational constraints, scope boundaries, iteration policies, completion criteria, and pause conditions. The skill prioritizes generating a conservative, directly executable default version of the goal, complete with explanations, and offers numbered options for customization, ensuring clarity and reducing ambiguity for the AI.

Quick Start & Requirements

  • Install: npx skills add joeseesun/qiaomu-goal-meta-skill
  • Prerequisites: An environment supporting Agent Skills (e.g., Codex, Claude Code), Node.js, and npx. The execution environment must be able to read from ~/.agents/skills.
  • Links: GitHub repository: https://github.com/joeseesun/

Highlighted Details

  • Generates a copy-ready, recommended execution version of the /goal command by default, eliminating the need for manual form-filling.
  • Includes comprehensive verification, constraints, boundaries, iteration policy, completion, and pause conditions tailored to the task.
  • Offers concise, numbered options for customizing key aspects like project form, visuals, scope, and verification methods.
  • Provides an English-compatible mirror of the goal fields for broader agent compatibility.
  • Features a lightweight linter (scripts/lint_goal_command.py) to validate generated /goal commands against common issues.

Maintenance & Community

The project is authored by 向阳乔木 (Vista8). While direct community links like Discord or Slack are not provided, the author's GitHub and X (formerly Twitter) profiles are available for contact.

Licensing & Compatibility

The project is released under the MIT License, permitting commercial use and modification.

Limitations & Caveats

This skill exclusively generates /goal commands; it does not execute the goals themselves. It is designed not to bypass restrictions related to copyright, accounts, payments, production data, or compliance. It defaults to generating Minimum Viable Product (MVP) level goals, requiring explicit user requests for full product strategy, commercialization, or deployment details. For unfamiliar professional domains, it defaults to a "discovery-first" approach rather than assuming expertise.

Health Check
Last Commit

1 week ago

Responsiveness

Inactive

Pull Requests (30d)
1
Issues (30d)
1
Star History
653 stars in the last 7 days

Explore Similar Projects

Feedback? Help us improve.