Two tools with basic schemas and descriptions, but significant gaps in parameter documentation, output schema clarity, and error handling guidance. Tool names follow verb_noun pattern (positive), but descriptions lack actionable context for LLM selection. Parameters have types but minimal guidance on constraints or expected formats. Error handling returns JSON strings but provides no recovery guidance. No tool annotations (readOnlyHint, etc.) despite both being read-only operations.

• Add formal output schema definitions to both tools. Document the structure of returned JSON (e.g. extract_and_analyze_smiles returns array of {smiles: string}, get_detailed_properties returns {smiles, formula, molecular_weight, ...}). Use JSON Schema format in tool registration.
• Enhance error messages with recovery guidance. Instead of 'Invalid SMILES string', return 'Invalid SMILES string. Ensure the string follows SMILES notation rules (e.g. C for carbon, O for oxygen, = for double bonds). Example valid SMILES: CCO (ethanol).'
• Add constraint descriptions to parameters. For 'text' parameter: 'Text containing zero or more SMILES strings wrapped in <smi>...</smi> tags. Example: "The compound <smi>CCO</smi> is ethanol."' For 'smiles' parameter: 'A valid SMILES string representing a chemical compound (e.g. CCO, c1ccccc1). Invalid SMILES will return an error.'
• Clarify tool selection context in descriptions. Add to extract_and_analyze_smiles: 'Use this to extract and validate SMILES strings from text. Call this first if you have raw text containing chemical notations.' Add to get_detailed_properties: 'Use this after extracting a SMILES string to retrieve detailed molecular properties (weight, formula, bond counts, etc.).'
• Add readOnlyHint annotations to both tools in FastMCP registration to signal they are safe to retry and have no side effects. This enables agents to optimize error recovery.