Is Pesky MCP safe to give an agent?

destructive capableremoteunverified

A factual risk summary built from Pesky’s real tool surface, execution model, and verification history — not a vibe.

01What it can do

Has tools that can delete or irreversibly change data.

15 tools observeddestructive presentwrite present

02Execution model

Runs on the vendor's infrastructure; you connect over the network. No untrusted code runs on your machine, but you grant the hosted service access.

Connects to a remote URL — no local package execution.

03Permissions & auth

OAuth 2.1 with two scopes: reading and scanning (mcp:read), or also changing competitor lists (mcp:write), chosen at sign-in. Or an API key, which acts as the account with no scopes.

04Verification

handshakenot run — tool list recorded from the vendor docs

runsnone

last_checked—

sourcesvendor docs

Reduce the risk

Worried about handing an agent raw access? See governed agents in action — Apex gives your AI paced, capped, fully-logged hands with approval queues before anything runs.

Explore Apex →

See also: full server page · setup · alternatives