Ranked #2 of 711/18 · Last reviewed

Built-in agent settings

The controls each agent ships with: permission modes, allow and deny rules for commands and tools, sandboxing, hooks, and for some agents managed settings an administrator can push.

Verdict: Necessary groundwork for one agent, not a program for many: use it, and manage it centrally.

Who it suits

Small teams that have standardized on one agent and can manage its configuration centrally.

Examples

Permission rules, hooks and managed settings in agents such as Claude Code, Cursor, Codex and Copilot

Cost

Included with the agent. Which admin and managed-settings features you get depends on the agent's plan: check each agent vendor's plans page.

Pairs with

An AASB to make the same rules org-wide, cross-agent and provable.

How it handles MCP connections
  • Discovery: Lists the MCP servers configured in that one agent.
  • Control: Allow or deny servers and tools by name in that agent's settings.
  • Record: Whatever the agent keeps locally; not central.

Strengths

  • Already in the tool: nothing to buy
  • Acts on commands and tool calls inside that agent
  • Managed settings can stop developers loosening the rules, where the agent and plan support them

Limits

  • Per agent: each tool has its own rules, syntax and gaps
  • Pattern rules (globs, prefixes) miss commands that do the same thing differently
  • No fleet inventory, and records stay with each agent
  • Approval prompts invite reflexive yes, or a switch to auto-approve

Scores, with reasons

Each criterion is scored 0 to 3 as defined on the criteria page.

CriterionScoreWhy
Visibility 1Each agent sees its own session; there is no view of which agents and MCP servers run across the fleet.
Enforcement 2Allow and deny rules and hooks act before a command or tool call runs, but only inside that agent and only as precise as its patterns.
Developer friction 2Strict settings prompt often; developers answer yes by reflex or switch to an auto-approve mode.
Coverage of terminal and MCP actions 2Covers terminal and MCP actions, but for one agent at a time.
Audit evidence 1Logs and transcripts stay with each agent and developer, in each tool's own format.
Deployment effort 3Already installed; managed settings can be pushed with existing device management where supported.
Total11/18Ranked #2 of 7. See the full table.
Other approaches