- Home
- Skills
- Agents & Orchestration
- mcp tool description engineer
Works with the AI tools you already use
mcp tool description engineer
Optimize MCP tool definitions and schemas to stop agents from calling the wrong tools or hallucinating parameters.
$11
mcp tool description engineer
Example session with this skill installed
My agent keeps using the 'get_log' tool for both system logs and application logs, but they require different parameters. How do I fix the tool definitions so it stops mixing them up?
- Read your context and instructions
- Compiled the mcp tool description
- Rename tools to 'get_system_audit_log' and 'get_app_runtime_log'.
- Add a WHEN clause: 'Use this only for OS-level events' vs 'Use for stack traces and app errors'.
- In inputSchema, change the generic 'id' parameter to 'service_name' with an enum of valid options.
Connects securely to your tools. The creator never sees your data.
What you get
About this skill
The problem
Model selection accuracy degrades as you add more tools to an MCP server. Agents call the wrong functions, skip required parameters, or hallucinate argument values because your tool definitions are treated as documentation rather than steering signals.
What it does
- Audits existing
tools/listpayloads for naming collisions, ambiguous parameter schemas, and missing trigger conditions. - Rewrites tool names and descriptions to emphasize the decision boundary between similar capabilities.
- Hardens
inputSchemadefinitions using enums, format hints, and descriptive parameter names to prevent hallucinated values. - Architects namespacing and grouping strategies for servers exceeding 15 to 20 tools to maintain selection reliability.
- Constructs a repeatable test battery to measure and verify tool-selection accuracy rates against realistic prompts.
Frameworks & tools
Model Context Protocol (MCP), JSON Schema, Anthropic tool-use specs.
Why this beats prompting it yourself
Writing tool definitions that "look right" to a human often fails in production because models respond to different linguistic cues. This skill applies verified patterns from primary MCP benchmarks to ensure the model picks the right tool every time, not just most of the time.
Use cases
- Fixing an agent that keeps calling
search_userswhen it needssearch_orders. - Refactoring a massive MCP server with 40+ tools that has become unreliable.
- Preventing parameter hallucinations by tightening loose
type: stringdefinitions. - Validating tool selection accuracy before shipping a new MCP server to production.
Known limitations
This tool focuses on functional correctness and selection accuracy. It does not perform security audits for prompt injection or tool-poisoning vulnerabilities.
How to install
Works the same in every agent - Claude, Cursor, Codex, Copilot and 20+ more.
- 1
Download the ZIP
Free skills download straight away. Paid skills unlock right after purchase.
- 2
Unzip into your skills folder
Every agent reads skills from one folder on your machine. Drop the unzipped folder in there.
- 3
Ask your agent to use it
Restart the agent if it was already running. It picks the skill up automatically - no config needed.
Skills folder by agent
Click the path to copy it. Create the folder if it does not exist yet.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill passes our trust checks below.
Security scanned
Passed our 8-point scan before listing
Fresh listing
Recently published to Agensi
30-day refund
Not a fit? Get your money back
Trust & safety
Security scanned
Verified clean 23 days ago
- Passed all security checks, Safe to install