Skip to main content Announcing Tool Policies: AI agent permissions down to the field Read the announcement
Action Effects in MCP Search Results

November 2026

Action Effects in MCP Search Results

Search results on the Gateway MCP endpoint now carry each action’s effects. Every action that stackone_search_actions returns lists the effects its connector declares for it, so an agent can see that an action writes or deletes before it calls it instead of guessing from its name.

What’s new

  • effects on every search hit - Each action in the result has an effects list made of read, search, write, delete and execute. The field is always present, and an action that declares no effects returns an empty list.
  • Both search tools - The session-wide stackone_search_actions tool and the search tool scoped to a single account both return the field.
  • Taken from the account’s connector version - Effects come from the connector version each matching account runs. When accounts on different versions match the same action, the result lists every effect any of those versions declares.
  • Synced records read as reads - A search hit that queries records kept by Data Sync carries ["read"].

Using effects for approval

With search and execute, an agent sees two tools rather than one tool per action, so per-tool MCP annotations do not apply to the actions it finds. The effects list on each search hit fills that gap: an agent or MCP client can run read and search actions straight away and ask the user before it executes a write or delete.

The values are the same ones behind the MCP tool annotations on individually exposed tools, and connector authors set them as described in the connector YAML reference. How agents find actions in the first place is covered in the Search and Execute entry.

Put your AI agents to work

Connect AI to every company system and control what it can do in each one, down to the field.