Agentvet
Validate tool-call arguments before execution and return retry hints the model can use to fix bad params.
Overview
Agentvet is an MCP server for the Ship phase that validates tool-call arguments before execution and supplies LLM-friendly retry hints.
What is this MCP server?
- Validates tool-call arguments prior to execution
- Returns LLM-friendly retry hints on validation failure
- stdio MCP @mukundakatta/agentvet-mcp on npm
- Shifts argument checking out of ad hoc try or catch in each tool
- Package @mukundakatta/agentvet-mcp version 0.1.0
- Transport type stdio
- Repository github.com/MukundaKatta/agentvet-mcp
Community signal: 1 GitHub stars.
What problem does it solve?
Invalid tool arguments slip through and cause failed runs without structured feedback for the model to self-correct.
Who is it for?
Builders with many MCP tools who want consistent argument validation and clearer retry loops.
Skip if: Tools with no structured args or teams that validate only on the server after damage is done.
What do I get? / Deliverables
Bad tool calls are blocked pre-execution with hints the agent can use on the next attempt.
- Validation pass or fail for tool arguments
- Structured retry guidance for the calling model
Recommended MCP Servers
Journey fit
How it compares
MCP pre-execution validator, not a post-hoc log analyzer like trace snapshot tools.
Common Questions / FAQ
Who is agentvet for?
Solo developers running multi-tool agents who need schema-style checks before side-effecting calls.
When should I use agentvet?
Use it when shipping or hardening agent flows where bad parameters cause failed deploys, edits, or API calls.
How do I add agentvet to my agent?
Install @mukundakatta/agentvet-mcp, register stdio MCP in your client, and invoke validation on proposed tool calls before execution.