edgar-mcp has 8 well-named tools with clear, domain-specific descriptions (avg 120 chars). All tools have input schemas with typed parameters and descriptions. However, output schemas are not documented in the visible code, only inferred from tool logic. Error handling is minimal; tools raise EdgarError but provide no recovery guidance. Parameter descriptions lack format constraints (e.g., date format YYYY-MM-DD is stated but not enforced via schema). No tool annotations (readOnlyHint, etc.). Composition is sound, tools chain naturally (resolve_company → list_filings → get_filing_text). Security is adequate for read-only operations but lacks audit logging.

• Document output schemas for all tools. Specify field names, types, and structure (e.g., lookup_company returns {cik: int, ticker: string|null, name: string}).
• Add recovery guidance to error scenarios. E.g., 'If company not found, try search_filings() with a partial name or ticker.'
• Add readOnlyHint: true to all tool schemas to signal to clients that these tools are safe to call without user confirmation.
• Add regex pattern or format constraint to date parameters: 'since' and 'until' should validate ^\d{4}-\d{2}-\d{2}$ in schema.
• Return total count and next_cursor in paginated results (list_filings, search_filings, get_concept, list_concepts, compare_concept) so LLMs know if more data exists.