abap-mcp provides 10 read-only tools with basic descriptions and input schemas. All tools are verb-prefixed (abap_*) and have non-empty descriptions (range 50 - 150 chars). However, parameter descriptions are minimal or absent, output schemas are undocumented, and error handling lacks recovery guidance. No tool annotations (readOnlyHint, idempotentHint) despite all being read-only. The server is well-intentioned but falls short of production-grade definition quality.

• Document output schemas for all tools. For abap_search, specify: {ok: boolean, count: integer, results: [{name: string, type: string, uri: string, package: string, description: string}]}. For abap_read_class, specify: {ok: boolean, source_uri: string, etag: string, source: string, object_uri: string}. This enables LLMs to chain tools and extract fields correctly.
• Add parameter constraints to descriptions. For abap_search 'max_results', write: 'Maximum number of results (1 - 100, default 20). Larger values may timeout.' For abap_sql 'select', write: 'Read-only SELECT statement only. No INSERT, UPDATE, DELETE, or DDL. Max 1000 rows returned.'
• Enhance error messages with recovery guidance. Instead of returning {ok: false, error: 'HTTP 401\n...'}, return {ok: false, error: 'Authentication failed (HTTP 401). Verify SAP_USER, SAP_PASSWD, and SAP_BASE_URL. Call abap_ping to test connectivity.'}. Categorize errors: retryable (timeout, 503), user-fixable (401, 404), fatal (500).
• Add tool annotations to tool definitions. Modify tool registration to include readOnlyHint: true for all tools and idempotentHint: true for abap_ping, abap_search, and all abap_read_* tools. This signals to LLMs that these tools are safe to retry and do not modify state.
• Implement pagination for abap_search. Add offset and limit parameters (default limit=20, max=100). Return {ok, count, total, offset, limit, results}. This prevents context window exhaustion and enables iterative discovery of large result sets.