// SCAN REPORT

Mcp

Q: Can an AI agent delete data through the MCP server?

Yes. The Mcp server exposes 4 destructive tools including delete_site, delete_variable, drive_delete_file. These permanently remove resources with no undo. PolicyLayer blocks destructive tools by default so they never reach the upstream server.

Q: How do I prevent bulk modifications through Mcp?

The Mcp server has 10 write tools including claim_site, create_drive, drive_move_file. Set a rate limit in your policy -- for example, 10 calls per hour prevents an agent from making more than 10 modifications per hour. PolicyLayer enforces this at the gateway, before calls reach Mcp.

Q: How many tools does the MCP server expose?

26 tools across 3 categories: Destructive, Read, Write. 12 are read-only. 14 can modify, create, or delete data.

Q: How do I enforce a policy on Mcp?

Register the MCP server in PolicyLayer, apply the suggested rules above (adjust the limits to your use case), and point your AI client at the PolicyLayer proxy URL instead of the server directly. Your agents keep the same tools; PolicyLayer evaluates every call against policy before it executes. Nothing to install, live in minutes.

26 tools. 14 can modify or destroy data without limits.

4 destructive tools with no built-in limits. Policy required.

Last updated: 9 Jul 2026

14 can modify or destroy data

12 read-only

26 tools total

Community server · catalogue entry checked 09/07/2026 · full schemas captured for 22 of 26 tools

How to control Mcp ↓

TOOL MAP

What Mcp exposes to your agents

Read (12) Write / Execute (10) Destructive / Financial (4)

Running more than one server? Check your whole stack →

MOST DANGEROUS TOOLS

Critical Risk

The most dangerous Mcp tools

Destructive · High delete_site This tool irreversibly removes a deployed site, destroying all associated data and configurations. Destructive · High delete_variable This tool deletes an account variable, which is an irreversible destructive action. Destructive · High drive_delete_file This tool permanently removes files from persistent storage (a drive used for agent memory). Destructive · High drive_revoke_token Revoking a share token is an irreversible action that permanently invalidates access credentials.

14 of Mcp's 26 tools can modify, destroy, or commit something on every call — and an agent calls them with no built-in limits.

POLICY

How to control Mcp

PolicyLayer is an MCP gateway — it sits between your AI agents and Mcp, and nothing reaches the server without passing your rules. These are the rules we recommend:

Deny destructive operations

{
  "delete_site": {
    "deny_if": [
      {
        "conditions": [],
        "on_deny": "Blocked by default. Requires approval."
      }
    ]
  }
}

Destructive tools should never be available to autonomous agents without human approval.

Rate limit write operations

{
  "claim_site": {
    "limits": [
      {
        "counter": "claim_site_per_hour",
        "window": "hour",
        "max": 30,
        "scope": "grant"
      }
    ]
  }
}

Prevents bulk unintended modifications from agents caught in loops.

Cap read operations

{
  "app_status": {
    "limits": [
      {
        "counter": "app_status_per_minute",
        "window": "minute",
        "max": 60,
        "scope": "grant"
      }
    ]
  }
}

Controls API costs and prevents retry loops from exhausting upstream rate limits.

Create a free account and register Mcp — nothing to install.
Add these rules — paste them, or build them visually. Tune the limits to your setup.
Point your MCP client (Claude, Cursor, anything) at your gateway URL.

ENFORCE POLICY ON MCP →

Instant setup, no code required.

FULL CATALOGUE

All 26 Mcp tools

DESTRUCTIVE 4 tools

Destructive delete_site Permanently delete an owned site. Irreversible — confirm with the user first. Destructive delete_variable Delete an account variable. Destructive drive_delete_file Delete one file from a drive. Destructive drive_revoke_token Revoke a drive share token.

WRITE 10 tools

Write claim_site Make a previously-anonymous site permanent under the authenticated account (requires auth + the claimToken fro Write create_drive Create a new private drive. Write drive_move_file Move/rename a file inside a drive. Write drive_restore Restore a drive to a prior version (commits a new version; nothing is lost). Write drive_share Mint a scoped drive access token for another agent (perms read|write, optional path prefix + TTL like 7d). The Write drive_write_file Write/overwrite one file in a drive (ETag-safe: pass ifMatch to assert the version you read, otherwise the cur Write publish_site Deploy files to a live URL. No slug → create a NEW site (works without auth; anonymous sites expire in 24h — a Write request_login_code Start sign-in: email a one-time code to the user. Follow with verify_login_code. Write set_variable Create/update a write-only account secret (uppercase name). Injected as env vars into container apps and as ${ Write update_site_metadata Patch site settings without redeploying: viewer title/description, SPA mode, TTL, password (string sets, null

READ 12 tools

Read app_status Build/run status of a container app deployed via publish_site with a Dockerfile. Poll until status is live or Read drive_history Version timeline of a drive (every write is a restorable version). Read drive_list_files List files in a drive (by drive name or drv_… id), optionally under a path prefix. Read drive_read_file Read a file from a drive. Text comes back as utf-8 content; binary as contentBase64. Files over 1 MB are refus Read drive_tokens List active share tokens of a drive (secrets are never shown again). Read get_site Full details + current file manifest of an owned site. Read list_drives List the authenticated account's private cloud drives. Read list_sites List all sites owned by the authenticated account. Read list_variables List account variable NAMES (values are write-only — they are injected into proxy calls and container apps, ne Read platform_info DataEcho platform overview: what it can host, auth state of this connection, docs links. Call this first if un Read search_sites Full-text search across the authenticated account's sites (names, metadata, and text content). Read verify_login_code Complete sign-in with the emailed code. Returns the apiKey ONCE — configure it as the Authorization: Bearer <k

EXPLORE

Related servers

Other MCP servers with similar tools — same risk classification, starter policies for each.

FAQ

Questions about Mcp

Can an AI agent delete data through the MCP server? +

Yes. The Mcp server exposes 4 destructive tools including delete_site, delete_variable, drive_delete_file. These permanently remove resources with no undo. PolicyLayer blocks destructive tools by default so they never reach the upstream server.

How do I prevent bulk modifications through Mcp? +

The Mcp server has 10 write tools including claim_site, create_drive, drive_move_file. Set a rate limit in your policy -- for example, 10 calls per hour prevents an agent from making more than 10 modifications per hour. PolicyLayer enforces this at the gateway, before calls reach Mcp.

How many tools does the MCP server expose? +

26 tools across 3 categories: Destructive, Read, Write. 12 are read-only. 14 can modify, create, or delete data.

How do I enforce a policy on Mcp? +

Register the MCP server in PolicyLayer, apply the suggested rules above (adjust the limits to your use case), and point your AI client at the PolicyLayer proxy URL instead of the server directly. Your agents keep the same tools; PolicyLayer evaluates every call against policy before it executes. Nothing to install, live in minutes.

Enforce policy on every Mcp tool call.

Deterministic rules across all 26 Mcp tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.

ENFORCE POLICY ON MCP →

Instant setup, no code required.

26 Mcp tools catalogued and risk-classified — across an index of 46,500+ MCP servers.

// WHERE THIS COMES FROM

These policies come from 's registry record.

The record behind this page: verified identity, auth posture, risk grade, every tool classified, recommended policy — re-checked continuously.

Teams ship this data inside their own products. See what a licence covers →