Elevator pitch Automated linter that scans MCP servers, detects vague tool descriptions and ambiguous parameters, then rewrites schemas to eliminate AI agent hallucination and task failure.
Industry AI tooling & MCP infrastructure.
Problem
- Vague tool descriptions cause agents to misuse or skip available functions.
- Ambiguous parameter constraints lead to hallucination, infinite loops, and failed tasks.
Solution
- Scans live MCP servers and identifies "tool smells" in real time.
- Auto-generates precise descriptions, constraints, and structural fixes.
Tools
SchemaScanner: Introspects MCP server tools, returns vagueness scores and ambiguity flags.DescriptionRewriter: Accepts tool metadata, returns clarified descriptions with examples.ConstraintInferrer: Analyzes parameter usage patterns, returns tightened enum/range constraints.SchemaValidator: Tests rewritten schemas against agent behavior logs, returns pass/fail report.
Widgets
/dashboard: Real-time server health, linting results, and rewrite recommendations./compare: Side-by-side before/after schema diffs with impact metrics.
Conversation starters
- "Scan my MCP server and show me which tool descriptions are confusing agents most."
- "Rewrite this parameter schema to eliminate ambiguity and add validation examples."