Multi-tool MCP server providing SEO content drafting, content generation, content summarization, and weather alerts via integration with FastMCP and external APIs
This server has critical gaps in definition quality across all dimensions. Only 2 of 5 tools have visible input schemas in the source code provided. Descriptions range from minimal (15-20 chars) to moderate, but lack the clarity and actionability needed for LLM-driven tool selection. Parameters are under-described, with no enum constraints, format specifications, or dependency documentation. Tool composition is problematic: get_alerts (weather) and echo_resource (echo) are misplaced in a SEO/Content server, creating semantic confusion. No output schemas are documented. Error handling is absent, tools return placeholder strings with no recovery guidance. The server lacks idempotency hints, permission scopes, and audit capabilities.
Echo a message as a resource
Generate content based on the specified type, length, and format.
Get weather alerts for a US state.
Draft content for SEO blog posts with specified tone
Summarize the provided content to a specified length.
Critical: Tool composition is incoherent. get_alerts (weather) and echo_resource (echo) belong in a weather server, not a SEO/Content server. This semantic mismatch will cause LLMs to misselect tools and fail at composition tasks.
Critical: Parameter descriptions are incomplete or missing. 'topic' (seo_blog_draft), 'tone' (seo_blog_draft), and 'content_format' (generate_content) have minimal 1-2 word descriptions. LLMs cannot infer whether 'tone' accepts 'professional|casual|technical' or free-form strings.
High: No enum constraints on enumerable parameters. 'tone' in seo_blog_draft and 'content_format' in generate_content should be enums (e.g., tone: ["professional", "casual", "technical"]), not free-form strings. Without enums, LLMs hallucinate invalid values.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 47 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 37 | - | v1 |
High: No output schemas documented. Tools return plain strings (e.g., 'Draft (...) on {topic}: ...') with no structure. LLMs need to know what fields to expect so they can parse and compose results.
High: Incomplete parameter descriptions. 'generate_content' documents only 'content_format' in its schema, but the function signature shows it accepts 'content_type' and 'content_length' as hardcoded locals (500 words, 'article'). The schema is missing these parameters entirely, making the tool unusable, LLMs cannot control the actual generation parameters.
High: No error handling or recovery guidance. All tools return placeholder strings on success but do not handle or describe failure modes. Per pattern:recovery-guide, error responses must tell the LLM what to do next. E.g., get_alerts returns 'Unable to fetch alerts...' but does not suggest retry parameters or alternative approaches.
Medium: Ambiguous tool naming. 'generate_content' is vague, does it draft blog posts, create marketing copy, generate product descriptions? The name alone does not convey the specific action. Baseline naming analysis shows verb_noun patterns (e.g., 'draft_blog_post', 'create_marketing_copy') improve LLM disambiguation. Current name is generic and overlaps semantically with seo_blog_draft.
Medium: 'echo_resource' has a nonsensical name for a messaging tool. 'Echo' is typically a debugging/test operation. The tool description says 'Echo a message as a resource', unclear what this actually does or when an LLM should call it. Is it for testing? For broadcasting? For storing? Name and description are misaligned.
Medium: Missing pagination guidance. 'get_alerts' may return many alerts for a state. No limit, pagination (page/offset/limit), or total_count documented. Per pattern:paginated-result, tools returning lists should accept pagination parameters and return a total count. Large results blow context windows.
Low: Tool descriptions lack context on WHEN to use them vs alternatives. E.g., seo_blog_draft vs generate_content, both produce content, but the distinction is unclear.