PlainGov provides 6 tools with complete JSON schemas and basic descriptions. All tools follow verb_noun naming (explain_, get_, eligibility_, generate_, timeline, questions_). Schemas are well-formed with enum constraints on program_id. However, descriptions are generic (10-50 chars), parameter descriptions are minimal or absent, output schemas are undocumented, and error handling lacks recovery guidance. No tool annotations (readOnlyHint, destructiveHint). The eligibility_check tool accepts a complex user_context object but provides no guidance on which fields are required vs optional or how missing data affects results. No pagination, no result limits documented. Typical of community servers in the 50-60 range.

• Expand tool descriptions to 100-200 chars. Example: 'explain_program: Retrieve a plain-language summary of a government program (e.g., GST Credit, Canada Child Benefit). Use this to understand program basics before checking eligibility or generating an application checklist.'
• Document output schemas for all tools. Add a 'Returns' section to each tool description or include a JSON example. E.g., explain_program returns {text: string, lastVerified: string, source: string}.
• Add parameter descriptions to eligibility_check.user_context. Specify which fields are required, which are optional, and how missing data affects the eligibility result (e.g., 'income: optional, number. If omitted, eligibility status will be unclear.').
• Add readOnlyHint: true to all 6 tool schemas to signal to LLMs that these are safe, non-destructive calls.
• Enhance error messages with recovery guidance. E.g., 'Retrieval failed (HTTP 404). The program information may be temporarily unavailable. Try a different program or contact support.'