MCP server that integrates with the Graphlit Platform, an LLM-enabled knowledge API platform supporting content ingestion, RAG pipelines, and conversation management across multiple data sources including Slack, Discord, websites, Notion, Google Drive, email, Jira, Linear, and GitHub.
The Graphlit MCP server defines 2 tools with visible schemas and descriptions. Tool naming follows verb_noun convention (configureProject, queryProjectUsage). However, queryProjectUsage lacks any visible description or schema in the provided source code, which is a critical gap. configureProject has a detailed description and proper Zod schema with parameter descriptions, but the overall quality is dragged down by the incomplete second tool and missing output schema documentation. The description for configureProject is verbose (200+ chars) but informative. No tool annotations (readOnlyHint, destructiveHint) are present. Error handling is basic, the code includes a switch statement validating modelServiceType but lacks comprehensive recovery guidance for LLM consumers.
Configures the default content workflow and conversation specification for the Graphlit project. Only needed if user asks to configure the project defaults. *Do not* call unless specifically asked for by the user. To reset the project configuration to 'factory state', assign False or null to all parameters. Optionally accepts whether to configure the default specification for LLM conversations. Defaults to using OpenAI GPT-4o, if not assigned. Optionally accepts whether to enable high-quality document and web page preparation using a vision LLM. Defaults to using Azure AI Document Intelligence for document preparation, if not assigned. Optionally accepts whether to enable entity extraction using LLM into the knowledge graph. Defaults to no entity extraction, if not assigned. Optionally accepts the preferred model provider service type, i.e. Anthropic, OpenAI, Google. Defaults to Anthropic if not provided. Returns the project identifier.
queryProjectUsage tool has NO visible description or input schema in source. Cannot assess parameter types, constraints, or LLM selection guidance.
No output schema documentation for either tool. LLMs cannot plan downstream calls or extract required fields (e.g., what fields does configureProject return? Is project_id included?).
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) present. configureProject is clearly a state-modifying tool (WRITE risk), but LLMs have no formal signal that it is destructive. queryProjectUsage is READ_ONLY but unmarked.
configureProject description is 200+ characters and somewhat verbose. While informative, it could be tightened to 80 - 150 chars for better token efficiency. Current baseline avg is 194, but A+ tools trend 50 - 200.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 56 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 43 | - | v1 |
Error handling in configureProject only covers modelServiceType validation (throws on unsupported type). No recovery guidance for LLM. E.g., if Graphlit API call fails, the error should indicate whether to retry, ask user, or abort.