MCP server for NetworkLytics — YouTube comment network analysis for AI agents
NetworkLytics MCP has 3 tools with explicit registration in server.py. Tool names follow verb_noun pattern (get_*), which is good. However, descriptions are present but somewhat generic, parameter descriptions are minimal, and output schemas are completely undocumented. The server reads structured API responses but does not guide LLMs on what fields to expect in the output. Error handling exists but provides only raw JSON error responses without recovery guidance. No input validation is documented, and no constraints (enums, ranges) are specified for parameters. Overall, this is a functional but below-average toolkit lacking the prompt-engineering rigor needed for reliable agentic use.
Retrieve a YouTube comment network analysis by its ID. Requires API key authentication (set NETWORKLYTICS_API_KEY environment variable). Returns the same structured data as get_shared_analysis.
Get NetworkLytics API information and available endpoints.
Retrieve a YouTube comment network analysis result from a public NetworkLytics share link. Returns network statistics (nodes, edges, density, communities), top influencers with centrality scores, sentiment analysis, topic keywords, and AI-generated insights. Use this when a user shares a NetworkLytics share URL or token.
No output schemas documented. Tools return raw JSON but do not tell LLMs what fields to expect in responses. This forces LLMs to infer structure from examples and risks misaligned downstream tool calls.
Parameter descriptions are minimal. 'token' is described but does not explain valid formats clearly (UUID vs full URL parsing is implicit). 'password' lacks context on when it is required. 'analysis_id' has no range or validation hint.
Error handling returns raw exception messages and API error bodies as unstructured JSON. No recovery guidance provided. LLMs receive 'error: HTTPStatusError' with no hint to retry, validate input, or call a different tool.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 56 | 2026-07-28+ | v2 |
No input validation or constraint documentation. The 'token' parameter accepts any string; no regex pattern or format constraint is declared. LLMs may pass invalid tokens without guidance on format.
Tool descriptions mention return types (e.g., 'network statistics, nodes, edges, density, communities') but do not document them formally. No field-by-field schema, so LLMs must guess at response structure.
get_api_info tool has minimal description ('Get NetworkLytics API information') and no guidance on when to call it or what structure it returns. This is a discovery tool but lacks discovery-oriented framing.