cli-over-mcp-auditor

cli-over-mcp-auditor is a skill for Claude Code from Abhillashjadhav/AI-PM-essential-skills. It costs 201 tokens per session (1,022 once invoked), scanned A, original, MIT.

A review tool for checking which connected tool servers are loaded in an agent session and whether they are relevant to the current task. MCP is a standard way for AI agents to connect to external tools.

In plain words
What is it for?
Use it to identify irrelevant connected tool servers and decide which ones could be disabled, either during the first prompt or in a deeper configuration audit.
Why use it?
Unused connections add tool descriptions to the agent's working context, which can make it harder to focus and consume available context space.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: mentions Claude Code.

Part of the pm-tactical plugin — 5 skills, 3 agents shipped together

Good fit Use it to identify irrelevant connected tool servers and decide which ones could be disabled, either during the first prompt or in a deeper configuration audit.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/abhillashjadhav/ai-pm-essential-skills/cli-over-mcp-auditor
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add Abhillashjadhav/AI-PM-essential-skills --skill cli-over-mcp-auditor
Clone the repo
git clone --depth 1 https://github.com/Abhillashjadhav/AI-PM-essential-skills

Made for: Claude Code.

Or install pm-tactical, the plugin that ships this one along with the rest of its 5 skills, 3 agents.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for cli-over-mcp-auditor

README.md
[![agentmods](https://agentmods.dev/badge/skills/abhillashjadhav/ai-pm-essential-skills/cli-over-mcp-auditor/github.svg)](https://agentmods.dev/skills/abhillashjadhav/ai-pm-essential-skills/cli-over-mcp-auditor)
Your own site
<a href="https://agentmods.dev/skills/abhillashjadhav/ai-pm-essential-skills/cli-over-mcp-auditor"><img src="https://agentmods.dev/badge/skills/abhillashjadhav/ai-pm-essential-skills/cli-over-mcp-auditor/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for cli-over-mcp-auditor

Your own site · 80×15
<a href="https://agentmods.dev/skills/abhillashjadhav/ai-pm-essential-skills/cli-over-mcp-auditor"><img src="https://agentmods.dev/badge/skills/abhillashjadhav/ai-pm-essential-skills/cli-over-mcp-auditor.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 201 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,022 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00201 $0.01022
Opus 5 $0.00101 $0.00511
Sonnet 5 $0.00040 $0.00204
Haiku 4.5 $0.00020 $0.00102

Measured 12d ago against content hash 43fdc6b56a46, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade A, and why

cli-over-mcp-auditor scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

For each server, check whether a mature CLI covers the same jobs (e.g., GitHub → `gh`, Vercel → `vercel`, cloud providers → their CLIs, databases → native clients, file/web jobs → curl or scripts). Rate replacement diffi
pm-tactical/skills/cli-over-mcp-auditor/SKILL.md · 63 lines

How it starts

The opening of the file, as written. The whole thing — 63 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLI-over-MCP Auditor

Every connected MCP server loads its tool definitions into the context window at session start — a standing tax paid whether or not the tools are used. CLI tools cost zero context until invoked. This skill finds the tax and recommends where to stop paying it.

Step 0 — First-prompt check (proactive, silent)

On the first substantial prompt of a session, before doing anything else, silently inspect the session's own already-available tools/MCP servers — do not ask the user to run or paste claude mcp list for this check; that's only for the explicit deep-audit flow (Step 1). Count connected MCP servers and their tools directly from what's visible in the session.

Judge relevance against the task the user is actually asking for right now — not against hypothetical future use. If one or more connectors look irrelevant to this task, emit exactly one line, then continue straight into the task:

N MCPs loaded, M unused for this work — disable to free up to ~X% context?

If nothing looks unused or irrelevant, say nothing — stay silent. This check runs once per session, not on every prompt.

Step 1 — Inventory (explicit deep audit)

List every connected MCP server. In Claude Code, ask the user to run claude mcp list and paste the output, or read .mcp.json / settings files if accessible. For each server, list its tools and count them.

Step 2 — Estimate idle cost

For each MCP server, estimate context cost as the token size of its injected tool definitions (tool names + descriptions + parameter schemas). Method, in order of preference:

  1. If the raw tool definitions are visible, count them directly (state the count as measured).
  2. Otherwise estimate: typical tool definition ≈ 150-600 tokens depending on schema complexity; multiply by tool count and label the result ESTIMATE with the range.

Never present an estimate as a measurement. Show per-server and total overhead as a percentage of a 200K context window.

Step 3 — Check usage vs cost

Read the full file on GitHub · 63 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 12d ago First seen · 63 lines · 201 tokens per session scan A 43fdc6b56a46

Subscribe to this mod's changes

cli-over-mcp-auditor is a skill published in the GitHub repository Abhillashjadhav/AI-PM-essential-skills (4 stars, last pushed 11d ago), licensed MIT. It adds 201 tokens to every session and 1,022 once invoked, about $0.0010 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.