A simple MCP server with packaged tools powered by Postman Runtime for OneDeck API integration
This server has critical definition quality issues across nearly all dimensions. Tool descriptions are present but extremely generic and lack actionable guidance. Most critically, ALL 12 tools have completely empty input schemas (type: object with no properties and empty required arrays), violating the fundamental requirement that parameters be properly typed and described. Tool naming is mixed, some follow verb_noun pattern (get_boards, post_create_record) but others are unclear (new_request, get_boardsboard_idfields appears to have a formatting error). The descriptions mention required environment variables as if they were tool parameters, which is fundamentally wrong, env vars should be injected server-side, not exposed to the LLM. No output schemas are documented. Error handling is basic (generic API error wrapping with no recovery guidance). No tool supports parameter constraints, enums, or format validation. The tool set appears to be a thin wrapper around an API (OneDeck) with minimal adaptation for agent usability.
Retrieves the list of all boards available in the system. Requires env vars: api_key, auth_apikey_in, auth_apikey_key, base_url
Fetches the list of fields for a specific board, including API key authentication. Requires env vars: api_key, auth_apikey_in, auth_apikey_key, base_url
Retrieves a specific document by ID, including API key authentication. Requires env vars: api_key, auth_apikey_in, auth_apikey_key, base_url, document_id
Fetches documents from the system, including API key authentication. Requires env vars: api_key, auth_apikey_in, auth_apikey_key, base_url
Retrieves details of a specific record within a board, including API key authentication. Requires env vars: api_key, auth_apikey_in, auth_apikey_key, base_url, board_id, record_id
Fetches all records from a specific board, including API key authentication. Requires env vars: api_key, auth_apikey_in, auth_apikey_key, base_url, board_id
ALL 12 tools have empty input schemas: type is 'object' with no properties defined and empty required arrays. Per HARD SCORING RULES, tools with no input schema must score 0 on schema. This is the most critical issue, agents cannot validate what parameters exist or what types they accept.
Tool descriptions list environment variables as prerequisites (e.g., 'Requires env vars: api_key, auth_apikey_in, auth_apikey_key') rather than actual tool parameters. Environment variables must be injected server-side via secret injection pattern. Exposing them as requirements in the description is a security anti-pattern and confuses the LLM about what is configurable.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 6 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 27 | - | v1 |
Retrieves a list of users from the system. Requires env vars: api_key, auth_apikey_in, auth_apikey_key, base_url
Retrieves general information from the API, including collection variables. Requires env vars: api_key, auth_apikey_in, auth_apikey_key
Creates a new record in a specific board, including API key authentication. Requires env vars: api_key, auth_apikey_in, auth_apikey_key, base_url, board_id
Sends an invite to share a studio item via shareId. Requires env vars: api_key, auth_apikey_in, auth_apikey_key, base_url
Uploads an attachment to a specific board record, including file details and MIME type. Requires env vars: api_key, auth_apikey_in, auth_apikey_key, base_url, board_id, file_name, mime_type, record_id
Updates a specific record within a board, including API key authentication. Requires env vars: api_key, auth_apikey_in, auth_apikey_key, base_url, board_id, record_id
Tool name 'get_boardsboard_idfields' appears malformed (missing underscore: should be 'get_boards_board_id_fields'). This makes the tool name ambiguous and hard to parse. Per naming baseline, clear verb_noun patterns are required for LLM intent inference.
Tool 'new_request' has a vague name that does not follow verb_noun pattern. 'new_request' is unclear about what it does, does it create, fetch, or initialize something? Compare to clearer names like 'get_api_info' or 'list_collections'. Per naming baselines, 90% of A+ tools start with action verbs like get, list, create, search.
No output schemas are documented for any tool. Agents cannot plan downstream calls or extract required chaining IDs (board_id, record_id, etc.) without knowing what fields are returned.
Tools get_users and post_invite both declare 'shareId' as a required parameter, but get_users is a READ_ONLY operation to retrieve users. This is inconsistent, a user listing endpoint should not require a shareId. Parameter relationships are undocumented.
Descriptions are generic and under 50 characters for most tools (e.g., 'Fetches documents from the system' is only 33 chars). These descriptions lack context for LLM selection, they do not explain WHEN to use the tool or WHAT it returns. Per hard scoring rule, descriptions under 20 chars score 0-20; most of these are marginal.
Error handling in mcpServer.js is basic: catches errors and wraps them as 'API error: {error.message}'. No error categorization (retryable vs user-fixable vs fatal), no recovery guidance, and no actionable suggestions. Per pattern:recovery-guide, errors must tell the LLM what to do next.
No pagination support visible in tool definitions. Tools like get_records, get_documents, get_boards return all results without limit, page, offset, or limit parameters. Per pattern:paginated-result and mxe:enforce-result-limits, large result sets blow context windows. The rubric requires pagination for list tools.
Destructive and write operations (post_create_record, put_update_record, post_upload_attachment, post_invite) have no dry-run, confirmation, or reversibility support. Per pattern:confirmation-request, irreversible operations should support a confirmation step to prevent agent mistakes.
Tool definitions do not indicate which are idempotent. Agents may retry on ambiguous failures, non-idempotent tools risk duplicate side effects (e.g., creating duplicate records). Per pattern:idempotent-operation, tools should declare idempotency to guide retry behavior.