Alongside a server’s Overview and Capabilities tabs, the server detail view has a Tests tab — a single place to see every capability with a saved test script and run them in bulk, without opening each tool/resource/prompt individually.

Layout

Capabilities are grouped the same way the Capabilities tab groups them — Tools, Resources, Prompts — with each group collapsed into its own card, showing only capabilities that have a saved script participating meaningfully (capabilities without one still appear, marked accordingly).

Per-capability status

Each row shows one of:

Status	Meaning
No test script	This capability has no saved script yet
Not tested	A script exists, but hasn’t been run this session
Running…	A run is currently in progress
Skipped — needs input	Auto-run couldn’t derive valid arguments (see below)
N passed / M failed	Result of the most recent run this session

Clicking an expandable row (one with a completed run) reveals the same per-test pass/fail rows used in the individual capability’s own Tests tab.

Running tests

• Run all tests — a single button runs every capability that has a saved script, using each one’s marked Saved Request (or schema defaults) to supply arguments.
• Per-row Run — a play icon next to any scripted row runs just that one capability.

Where arguments come from for an automated run

Since a bulk run has no live form to pull values from, MCPFlo derives arguments automatically:

1. Start from the tool’s/prompt’s schema defaults.
2. Overlay the capability’s Saved Request marked “use for tests”, if one exists (see Saved Requests) — the marked request’s values win per field; schema defaults still fill in anything it doesn’t cover.
3. Check that every required field now has a value.

If required fields still can’t be satisfied this way, the capability is reported “Skipped — needs input” rather than invoked with guessed or missing values — hovering the status explains: “Save a request on its Saved tab and mark it ‘Use for tests’.”

Aggregate summary

Once at least one capability has a completed run this session, the tab header shows a running total — N tests, passed, failed — aggregated across every capability on this server that’s finished a run. It stays hidden (not “0 of 0”) until the first run completes, so it never misrepresents an unstarted state as a clean pass.

Related

• See Saved Requests for how a request gets marked to feed automated runs.
• See Running Tests per Capability for the underlying per-capability Tests tab and result rendering this reuses.
• See Custom Tests Folder for where the scripts these rows check for actually live.