Codex/GitHub/Issue/26/02 Skills with implicit invocation disabled while only allowing explicit invocation

• Skills with implicit invocation disabled while only allowing explicit invocation #10585
  • body

    • What feature would you like to see?

      
      This issue is preceded by this question:
      

    • Specify that a skill cannot be invoked implicitly #10573

      ---

      
      Currently, there is no way to disable implicit invocation of skills while allowing explicit invocation via $<name>; we can only instruct codex in natural languages not to invoke skills implicitly, which may or may not be respected by codex.
      
      Given that custom prompts, which is similar to skills but can only be invoked explicitly (via /prompts:<name>), is deprecated in favor of skills, there should be a way to define skills with implicit invocations disabled; otherwise skills cannot be a complete alternative to custom prompts.
      

  • your-diary on Feb 3
    • to Author @etraut-openai

      It's a theoretical concern.
      

    • I have many custom prompts which perform destructive operations such as deleting AWS resources, so I want to be 100% (not 99%) sure they are invoked only when I explicitly request so.
  • wsmoak on Feb 14
    • Some tools use disable-model-invocation: true in the frontmatter for this -- then the skill will not be auto-invoked, only if you specifically ask for it. I thought it was in the agent skills spec but apparently not.
    • Cursor defines it like this:

    • By default, skills are automatically applied when the agent determines they are relevant. Set disable-model-invocation: true to make a skill behave like a traditional slash command, where it is only included in context when you explicitly type /skill-name in chat.
      

  • lattwood on Mar 23
    • One potential use-case for this functionality are skills that the agent might not be able to invoke in all environments the repository is being modified in.
    • Good point
      • For instance, an infrastructure type repo could have a skill for planning/previewing the changes (ie Terraform/Plan), but it requires SRE type credentials to be discoverable in the environment. This is not a skill you would want the model to automatically try to invoke if it's running somewhere like Codex Cloud, but it is a skill you would want to manually invoke from Codex CLI running on your laptop.
  • etraut-openai on Mar 30
    • This feature is already supported. See Codex/Skill/Optional metadata for details.
  • ignatremizov on Apr 3
    • For anyone landing here from search: this is configured in the skill metadata file agents/openai.yaml, not in SKILL.md.
    • Example:

    • policy:
        allow_implicit_invocation: false

    • Typical layout:

    • my-skill/
      ├── SKILL.md
      └── agents/
        └── openai.yaml

    • With that setting, the skill is not injected implicitly, but it can still be invoked explicitly via $skill-name.
  • MichaelYochpaz last week
    • I'd love it if Codex just supported the disable-model-invocation Frontmatter flag, as it's somewhat conventional and would work across other harnesses (from what I've found, this flag is currently supported on Claude Code, Gemini CLI, Cursor and Pi).
    • Good point
      • Forcing a custom openai.yaml file convention and making it Codex-specific is as-annoying as Anthropic not adopting the AGENTS.md file convention on Claude Code and forcing everyone to symlink CLAUDE.md.