mcp-tool

mcp-tool is a skill for Claude Code, Codex from skillberry-ai/cap-evolve. It costs 0 tokens per session (2,081 once invoked), scanned A, original, Apache-2.0.

A skill for improving how an agent uses an external MCP server's tools. It covers tool descriptions, parameter guidance, examples, and which tools are shown to the agent.

In plain words
What is it for?
Use it to refine an external server's tool presentation and documentation; it does not change the server's input schema or implementation.
Why use it?
It addresses agents choosing the wrong tools, supplying incorrect arguments, or being overwhelmed by a large tool list.

Skill for Claude CodeCodex

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.

agentmods
npx agentmods add skills/skillberry-ai/cap-evolve/mcp-tool
Any agent
npx skills add skillberry-ai/cap-evolve --skill mcp-tool
Clone the repo
git clone --depth 1 https://github.com/skillberry-ai/cap-evolve

Made for: Claude Code, Codex.

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 mcp-tool

README.md
[![agentmods](https://agentmods.dev/badge/skills/skillberry-ai/cap-evolve/mcp-tool.svg)](https://agentmods.dev/skills/skillberry-ai/cap-evolve/mcp-tool)
Your own site
<a href="https://agentmods.dev/skills/skillberry-ai/cap-evolve/mcp-tool"><img src="https://agentmods.dev/badge/skills/skillberry-ai/cap-evolve/mcp-tool.svg" alt="Measured on agentmods" height="20"></a>
Per session 0 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,081 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00000 $0.02081
Opus 5 $0.00000 $0.01040
Sonnet 5 $0.00000 $0.00416
Haiku 4.5 $0.00000 $0.00208

Measured 5d ago against content hash a0a585576029, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

mcp-tool scanned grade A with 0 findings 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 5d ago.

The scan reads SKILL.md. This mod also ships 4 executable files (scripts/_bootstrap.py, scripts/abstract.py, scripts/check.py, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/capabilities/mcp-tool/SKILL.md · 164 lines

How it starts

The opening of the file, as written. The whole thing — 164 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Capability: MCP tool (external server)

Tools served over the Model Context Protocol come from a server the agent does not own. The server defines each tool's name, description, and inputSchema and implements the handler; the host/client discovers them via tools/list, chooses which to present to the model, and invokes them via tools/call. So an optimizer here can change how the agent perceives and is offered those tools, and nothing else.

If the fix needs a tool's types, required fields, or behavior to change, this capability has been outgrown: negotiate the change with the server owner, or move the logic into an agent-owned tool and optimize that with the tools capability instead.

The edit boundary — read this before proposing anything

Edit Owner Allowed here
tool description, per-parameter description, in-description examples client presentation yes
which of the server's tools the model sees (add / remove) client/host curation yes
the wire inputSchematype, required, enum, maximum, properties (schema) server no
the handler implementation (code) server no
a new tool that runs server-side logic (compose) server no

The reason is not politeness: a candidate carrying a schema or handler edit is invalid against the real server. It cannot be deployed, it will fail at tools/call, and the run still pays full rollout cost to score it. The safe levers below are the whole edit space.

The policy checks the edit's LABEL, not its effect — so stay inside the boundary deliberately. apply() refuses any edit whose kind is outside the policy and reports the refusal, so a {"kind": "schema"} or {"kind": "code"} edit comes back as a visible refusal rather than a silent no-op. But two allowed kinds can carry a forbidden change through:

  • a params value is shallow-merged into parameters, so a value containing properties, type, required, or enum rewrites the wire schema and is not refused. Write params values that touch only properties.<field>.description.
  • an add value is appended verbatim, so a code key on it lands unrefused. add means expose a tool the server already serves; never invent one.

Read the full file on GitHub · 164 lines

Files

What ships with it

6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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. 5d ago First seen · 164 lines · 0 tokens per session scan A a0a585576029

Subscribe to this mod's changes

mcp-tool is a skill published in the GitHub repository skillberry-ai/cap-evolve (50 stars, last pushed today), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 2,081 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.