disable_model_invocation
When true, the AI cannot auto-invoke this skill; only the user can invoke it explicitly. Default: false.
Type: bool Content type: skills
Provider Support
Section titled “Provider Support”| Provider | Supported | Mechanism |
|---|---|---|
| claude-code | ✓ | yaml frontmatter key: disable-model-invocation (optional bool, default false); also prevents the skill from being preloaded into subagents via the skills frontmatter field |
| codex | ✓ | openai.yaml companion file key: policy.allow_implicit_invocation (optional bool; false disables implicit/auto invocation; defaults to true when absent) |
| crush | ✓ | yaml frontmatter key: disable-model-invocation (optional boolean) per the Agent Skills standard. Note: the Go Skill struct in skills.go does not have a corresponding field — behavior is inferred from the standard, not from explicit Crush source handling. |
| cursor | ✓ | yaml frontmatter key: disable-model-invocation (optional bool, default false); when true, the skill is converted into an explicit slash command rather than a context-aware tool the model can auto-invoke. |
| factory-droid | ✓ | yaml frontmatter key: disable-model-invocation (optional bool, default: false) |
| pi | ✓ | yaml frontmatter key: disable-model-invocation (optional bool); confirmed in SkillFrontmatter TypeScript interface as disable-model-invocation?: boolean |
| amp | ✗ | not documented — no mechanism to prevent the model from auto-invoking a skill; the model always selects based on name and description match |
| cline | ✗ | |
| copilot-cli | ✗ | |
| gemini-cli | ✗ | |
| kiro | ✗ | |
| opencode | ✗ | not documented |
| roo-code | ✗ | No disable-model-invocation frontmatter field observed |
| windsurf | ✗ | not documented |
| zed | ✗ | not documented |