Three tools with basic schemas and minimal descriptions. All tools have input schemas with typed parameters and descriptions, meeting baseline requirements. However, descriptions are extremely brief (18-24 chars), well below the 50-200 char LLM-optimized range. No output schemas documented. No error handling guidance. Tool names lack clear action verbs (impute, outliers, correlation are nouns/adjectives, not verb_noun patterns). Parameters use generic names (data_name, col_name) that don't follow the suffix-with-type convention. No indication of idempotency, side effects, or recovery paths. The impute tool modifies global state but lacks warnings about irreversibility.

• Expand tool descriptions to 80-150 chars. Example: 'impute' → 'Imputes missing values in a data frame column using the median. Modifies the data frame in place. Use when you have missing data and want to fill gaps with a robust central tendency measure.'
• Rename tools to verb_noun pattern: impute → impute_missing_values, outliers → detect_outliers, correlation → calculate_correlation.
• Document output schemas. Example for impute: {type: 'object', properties: {success: {type: 'boolean'}, message: {type: 'string'}, imputed_value: {type: 'number'}, rows_affected: {type: 'integer'}}}.
• Add error handling guidance. Example: 'If column not found, returns error listing available columns. If all values are NA, returns error with guidance to check data quality.'
• Rename parameters to be more explicit: data_name → data_frame_name, col_name → column_name, var1 → variable1_name, var2 → variable2_name.