Say a tool's hints in its description
Make a tool's description say in words what its read-only, destructive, idempotent and open-world hints say
Make a tool's description say in words what its read-only, destructive, idempotent and open-world hints say
Give tool (an MCP tool entry with description and annotations: readOnlyHint, destructiveHint, idempotentHint, openWorldHint). Returns the same tool with one sentence added after its description for each matching rule (reads only; confirm with the person before removing; adds or sets only; reaches outside), in rule order. A sentence the description already says is not added again, and a tool with no annotations comes back as it is.
Nothing: your AI starts it with your words.
Connect the AI you already use to Trillion once. Then ask in your own words; your AI finds this tool by what it does and runs it.
Say to your AI: “Use Trillion to make a tool's description say in words what its read-only, destructive, idempotent and open-world hints say”