Agent Utility MCP

by insivotron · remote · no sign-in to list tools · v0.3.0 · published by GitHub user insivotron

Stop your agent before it runs rm -rf /etc or emails your .env. Deterministic preflight checks.

Final · 1Changes · 1

Compare Registry entry

4 tools: 1 do something that can't be undone, 1 change things and 2 only read.

Scanned Oct 21, 2026

Deletes or cancels0Claude Code: deny
Can't be undone1Claude Code: ask: see these tools
Changes things1Claude Code: ask: see these tools
Only reads2Claude Code: allow: see these tools

2 checks matched something in the tool definitions. No finding is not a safety guarantee. How the checks work.

claude mcp add --transport http agent-utility https://agent-utility-mcp.insivotron.workers.dev/mcp

Findings

Matches of fixed checks, with the text they matched.

2 warn

Warninspect_command
The server marks "inspect_command" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
Matched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":true,"openWorldHint":false}; The name "inspect_command" says "inspect": a word for reading.
What this check looks for
Warnanalyze_tool_call
The server marks "analyze_tool_call" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
Matched text
annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":true,"openWorldHint":false}; The name "analyze_tool_call" says "analyze": a word for reading.
What this check looks for

Tools

Open a tool for its reasons and full description. Descriptions are the publisher's own words, shown as they are.

4 tools

Can't be undone 1Sends, pays, publishes or merges: once done, it cannot be taken back.Claude Code: ask

analyze_tool_callAnalyze a proposed AI-agent tool call before execution and return a deterministic, policy-aware decision (allow, notice, confirm or block) covering destructive actions, sensitive-d…1 finding
Why it is in this group
- The name "analyze_tool_call" says "analyze": a word for reading.
- The description says "irreversible".
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
  The server marks "analyze_tool_call" read-only (readOnlyHint: true), and its own name, description or input give the class "irreversible".
  Matched text
  annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":true,"openWorldHint":false}; The name "analyze_tool_call" says "analyze": a word for reading.
  What this check looks for
Description, as the server wrote it
Analyze a proposed AI-agent tool call before execution and return a deterministic, policy-aware decision (allow, notice, confirm or block) covering destructive actions, sensitive-data exposure, external transmission, privilege changes and irreversible operations, under the applicable policy — permissive, balanced or strict, balanced by default. Filesystem paths outside a known workspace_root are treated as higher risk than paths inside it.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false

Changes things 1Creates or changes something that can be changed back.Claude Code: ask

inspect_commandAnalyze a shell command before execution and return a deterministic, policy-aware decision (allow, notice, confirm or block) without running it, under the applicable policy — permi…1 finding
Why it is in this group
- The name "inspect_command" says "inspect": a word for reading.
- The input takes a statement or command to run (input.command), which can change anything it can reach.
- The server marks it read-only (readOnlyHint: true), but an annotation never lowers a class.
Findings
- Warn
  The server marks "inspect_command" read-only (readOnlyHint: true), and its own name, description or input give the class "write".
  Matched text
  annotations: {"readOnlyHint":true,"destructiveHint":false,"idempotentHint":true,"openWorldHint":false}; The name "inspect_command" says "inspect": a word for reading.
  What this check looks for
Description, as the server wrote it
Analyze a shell command before execution and return a deterministic, policy-aware decision (allow, notice, confirm or block) without running it, under the applicable policy — permissive, balanced or strict, balanced by default. Paths outside a known workspace_root are treated as higher risk than paths inside it.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false

Only reads 2Looks things up and changes nothing.Claude Code: allow, ask for 1 outbound

analyze_url_riskAnalyze a URL for structural and security risk signals and return a deterministic, policy-aware decision (allow, notice, confirm or block) under the applicable policy — permissive,…Reads text others writeCan send data outNo findings
Why it is in this group
- The name "analyze_url_risk" says "analyze": a word for reading.
What it can reach
- untrusted-inputThe name "analyze_url_risk" says "analyze" and "url": it reads web pages, which anyone can write.
- untrusted-inputThe description's first sentence says "analyze" and "url": it reads web pages, which anyone can write.
- untrusted-inputThe input takes a URL (input.url), and the tool reads it: it reads web pages, which anyone can write.
- outboundThe input takes a URL (input.url): a request to a URL can carry data to whoever runs it.
Claude Code: ask, because it is tagged outbound.
Description, as the server wrote it
Analyze a URL for structural and security risk signals and return a deterministic, policy-aware decision (allow, notice, confirm or block) under the applicable policy — permissive, balanced or strict, balanced by default.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false
inspect_fileInspect a Base64-encoded file locally for deterministic structural and security risk signals without executing it, and return a policy-aware decision (allow, notice, confirm or blo…Reads private dataNo findings
Why it is in this group
- The name "inspect_file" says "inspect": a word for reading.
What it can reach
- private-dataThe name "inspect_file" says "inspect" and "file": it reads files.
- private-dataThe description's first sentence says "inspect" and "file": it reads files.
Description, as the server wrote it
Inspect a Base64-encoded file locally for deterministic structural and security risk signals without executing it, and return a policy-aware decision (allow, notice, confirm or block) under the applicable policy — permissive, balanced or strict, balanced by default.
Annotations the server declares
- readOnlyHint: true
- destructiveHint: false
- idempotentHint: true
- openWorldHint: false

This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.

Exfil surface

Which tools read private data, which read text someone else can write, and which can send data out. One of each is a path a prompt injection can use.

1 path in this server alone from private data, through text someone else wrote, to a way out. Denying 1 outbound tool breaks every one.

Reads private data1: Reads files, mail, messages, databases, secrets, calendars or repositories.; inspect_file
Reads text others write1: Brings in text someone else can write: web pages, inbound email, messages, issues and comments.; analyze_url_risk
Can send data out1: Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.; analyze_url_risk

deny to break every path

{
  "permissions": {
    "deny": [
      "mcp__agent-utility__analyze_url_risk"
    ]
  }
}

Other servers in the same agent add their own tools to these paths. Audit your whole setup to see them together.

Install

Pinned to version 0.3.0. Values in angle brackets are yours to fill in.

claude mcp add --transport http agent-utility https://agent-utility-mcp.insivotron.workers.dev/mcp

Run in a terminal.

Claude Code permissions

Allow the reads, ask before a change or a way out, deny the deletes. Paste into .claude/settings.json.

.claude/settings.json

{
  "permissions": {
    "allow": [
      "mcp__agent-utility__inspect_file"
    ],
    "ask": [
      "mcp__agent-utility__analyze_url_risk",
      "mcp__agent-utility__inspect_command",
      "mcp__agent-utility__analyze_tool_call"
    ],
    "deny": []
  }
}

Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. A read that can send data out (it takes a URL, say) asks first too. It assumes the server is installed as agent-utility. Review it before you use it.

Capability tags behind 2 rules

mcp__agent-utility__analyze_url_risk untrusted-input, outbound
mcp__agent-utility__inspect_file private-data

Keep track

Get told when this server's tools change: a new tool, a new description, a new class.

https://anlyon.dev/servers/io.github.insivotron/agent-utility-mcp/changes.xml

An Atom feed. Works in any feed reader or Slack's RSS app.

Anlyon alternative

2 of these tools change something. Run those as governed actions.

Declare what an action changes, and Anlyon puts it under a shared limit, binds every approval to the exact request, and grades the outcome. It works on any HTTPS API you can describe as an action.

Anlyon does not run, proxy or govern this MCP server. A tool your own code calls directly is not routed through Anlyon and is not governed by it.

Import an OpenAPI spec →

Compare with other servers for the same product.

Source: MCP Registry. Synced Oct 21, 2026, 18:36 UTC.

For agents: Markdown · JSON · report this listing