MCP Server: io.github.alphaparkinc/genpark-levenshtein-fuzzy-tool-matcher-skill
This MCP server provides a fast Levenshtein distance and fuzzy similarity matcher for autocorrecting hallucinated LLM tool calls. It is designed to compare strings using Levenshtein-based similarity to identify intended tool names when generated tool calls do not match expected inputs.
๐ ๏ธ Key Features
- Fast Levenshtein distance computation
- Fuzzy similarity matching
- Autocorrection of hallucinated LLM tool calls
๐ Use Cases
- Correcting LLM-produced tool call names that do not match available tools
- Matching near-miss tool identifiers using edit-distance similarity
โก Developer Benefits
- More reliable tool call handling when tool names are misspelled
- Deterministic matching based on Levenshtein distance and fuzzy similarity
โ ๏ธ Limitations
- Scope described only for tool-call autocorrection; no additional capabilities are provided in the available data