Skip to main content
Tool definitions cost context on every request, and an AI client’s tool selection gets worse as the count climbs. Passport controls how the workspace endpoint presents tools so a large fleet stays usable. The inventory of what a person can reach stays visible at every tier; only the full tool definitions become lazy. You set the mode in Settings. The default is auto.

​
The three tiers

Flat

Every permitted tool is defined up front. Best experience with no indirection, for small setups.

Apps

One meta-tool per MCP, named for the MCP (for example github, supabase). The tool list itself is the readable inventory; each MCP’s tool definitions load on demand inside its meta-tool.

Compact

Two global meta-tools (a search and a call) whose descriptions carry the MCP manifest. For very large fleets.

​
How auto chooses

When the mode is auto, the workspace endpoint picks a tier from the count of permitted tools and MCPs for the caller:
1

Flat, until the threshold

While the total number of permitted tools is at or below the discovery threshold (default 40), the gateway serves flat.
2

Apps, past the threshold

Once the tool count passes the threshold, the gateway switches to one meta-tool per MCP, as long as there are 24 or fewer permitted MCPs.
3

Compact, for very large fleets

With more than 24 permitted MCPs, the gateway switches to the two global meta-tools.
You can pin any single tier instead of auto, and you can change the threshold (any value from 1 to 500).
Auto only applies to the workspace endpoint (/mcp and /w/{workspace}/mcp). A per-MCP endpoint (/s/{appKey}/mcp) and a bundle endpoint (/b/{bundleKey}/mcp) are already curated slices, so they always stay flat regardless of this setting.

​
Why the inventory always stays visible

Hiding the inventory would make an AI client capability-blind. It would answer “I don’t have GitHub” on clients that never proactively search. So at every tier the list of MCPs a person can reach stays readable. In apps mode the tool list is that inventory. In compact mode the meta-tool descriptions carry it. Only the detailed per-tool definitions load on demand, which is what keeps the context cost down. The repeatable context-efficiency benchmark measures the definition, task-context, wire, result, CLI-process, and latency costs of flat MCP, compact MCP, and Passport CLI against the same deterministic governed task.