There is a way to control tool invocations at the harness level when writing skills or agents.
Example: ~/.claude/agents/critic.md
--- name: critic description: > Plan Critic. Reviews an implementation plan. Use before the implementation. tools: Read, Agent ---
allowed-tools: Read Grep disallowed-tools: WebSearch Glob
There is a way to control tool invocations at the harness level when writing skills or agents.
Example: ~/.claude/agents/critic.md
For Claude skills, the frontmatter is different, the keys are: https://code.claude.com/docs/en/tools-reference