Guided first use
Try mcp-guard command protection
Reviews mcp-guard runs that start processes, approve a trust baseline or write message bodies to disk. proxy, test-injection --run and --liv Follow the three steps below with your own setup — nothing on this page runs on your machine or changes a policy.
Step 1 · Check fit
What this coverage governs
Reviews mcp-guard runs that start processes, approve a trust baseline or write message bodies to disk. proxy, test-injection --run and --live (inspect, lock, ui, snapshot save) start the MCP servers named in a config or on the command line, or the given agent script. lock writes the lock file that later runs compare against, so running it approves the current tools, including a changed description. proxy --record-bodies keeps full JSON-RPC messages, which can contain secrets, in the journal. Static audits (inspect, scan, ui, snapshot save and diff without --live), journal export, test-injection without --run, and direct runs with --help or --version (except proxy, whose later arguments go to the server) are not matched by this extension; Guard's default handling for commands no rule matches still applies to them. Covers the mcp-guard and mcp-guard-audit executables (including .cmd/.exe). uvx, uv tool run and pipx run launches of the package are covered by command.package.python.
External — off until you opt in
This coverage stays off until you explicitly enable it from your Guard workspace controls. Reading this page never changes your policy; activation happens in Guard itself.
Harness support
No harness compatibility profile is published yet. Check the Guard harness guide for what your setup uses.
Are you the publisher? Complete this section in your Studio — until then this stays unknown for readers instead of being guessed.
Catalog version 1.0.0 · source snapshot 54b61c819e
Step 2 · Set up
Install and activate
No reviewed setup steps are published for this coverage yet. Install Guard itself, then return here after the publisher publishes a guide — or ask for one. Are you the publisher? Complete this section in your Studio.
New to Guard? Start with Install Guard. Already installed? Skip install — this coverage stays off until you enable it from your workspace controls.
Activation happens in your Guard workspace. This page links to it; it never enables the coverage for you.
Missing the setup information you need? Request setup documentation — the publisher reviews requests before anything is published here.
Step 3 · Verify the result
Run a safe test and compare
Safe fixture inputs
No published safe fixture yet. Try the coverage against a harmless command of your own — not production traffic.
Are you the publisher? Complete this section in your Studio — until then this stays unknown for readers instead of being guessed.
Expected outcome
No expected outcome published yet. Guard surfaces what it matched — the rule and its decision — in its own activity view; compare that with this coverage’s stated rules to inspect the actual decision.
Are you the publisher? Complete this section in your Studio — until then this stays unknown for readers instead of being guessed.
Outcome reporting
This coverage has no claimed publisher page yet, so outcome reports are not collected. You can still verify the result locally against the expected outcome above.