AI Agent Hub
Back to skills
🤖

Skill Forge

AI Agent Updated 2026.08.30

Paste the following prompt into your AI chat to install this skill:

Please follow https://skillhub.cn/install/skillhub.md and install @user_164f4c1f/cjg-skill-forge.

About this skill

Problem

Agent skills often stall between a rough v1 and a releasable state: descriptions are too vague, triggers are unstable, scripts only “seem to work” in the author’s environment, skill libraries grow overlapping, and post-release feedback is hard to act on. cjg-skill-forge turns these problems into an engineering process, using real calls, evidence artifacts, and validation outputs instead of vague optimization advice.

How It Works

  • Forge mode: follows a S0–S8 pipeline covering scaffolding, real wiring, runtime evidence, external benchmarking, coverage audit, production sign-off, validation, packaging, clarity, and discoverability metadata.
  • Review mode: audits a skill against D1–D10 dimensions before release, surfaces anti-patterns, and produces a checkable report.
  • Recast mode: scans existing skills for overlap and consolidates similar items so the library does not keep fragmenting.
  • Clarity mode: rewrites skill content into a structure that AI can parse more reliably, with a fidelity guard to preserve intended behavior.
    It has clear limits: it does not make creative decisions, choose version numbers, or publish automatically; if a path has not been runtime-tested, it should be marked as non-runtime-tested rather than invented.

Use Cases

  • When creating a new skill, run the forge pipeline to add scaffolding, real wiring, runtime evidence, and release checks.
  • Before release, audit a skill against ten dimensions to catch weak triggers, missing commands, and coverage gaps.
  • When a skill library has duplicates, scan overlap, confirm consolidation, and tidy reusable documentation.
  • When vague docs confuse the model, rewrite the skill into a clearer structure while preserving intended behavior.

Best For

  • Agent skill-maintenance engineers who need to consolidate similar skills into a stable, maintainable structure.
  • Developers accountable for release quality who need to audit triggers, command syntax, and coverage gaps before launch.
  • Product engineers writing skill docs who want vague descriptions rewritten into clear, model-parseable structure.
  • Owners managing multi-version Agent skills who need to track usage signals and locate modules that need recasting.