Skip to content

自用/精选的 Claude Code / Codex 技能:内容创作、客服、市场调研、SEO、Agent 工作流循环、Grok X 实时抓取与评审工具

Python
1.1 KB
Raw Blame
interface:
  display_name: "Loopforge"
  short_description: "Compile, validate, repair, and package reusable AI-agent workflow loop specs."
  default_prompt: "Use $loopforge to turn this workflow, prompt, transcript, or artifact into a reusable loop spec. Distinguish it from a goal, SOP, checklist, cron job, one-shot task, or systems-thinking feedback loop. For YAML/JSON specs, also include a copy-ready Loop Prompt unless I ask for machine-only output. Lint specs before marking ready."
compatibility:
  canonical_format: "agent-skills"
  adapter_targets:
    - "openai"
    - "claude"
    - "generic"
  activation:
    mode: "manual"
    paths: []
  execution:
    context: "inline"
    shell: "bash"
  trust:
    source_tier: "local"
    remote_inline_execution: "forbid"
    remote_metadata_policy: "allow-metadata-only"
  degradation:
    openai: "creates and repairs loop-spec-v1 workflow specs from available task context"
    claude: "uses the same loop boundaries and spec fields with Claude-compatible prose"
    generic: "uses the canonical loop template and asks for missing high-impact criteria"