Cybersecurity code analysis API using FastAPI, OpenAI Agents, and Semgrep. Provides endpoints to analyze Python code for security vulnerabilities and return structured reports.
This MCP server has severe definition quality issues. Only 1 of 3 tools has a properly visible schema in the source code. The semgrep_scan tool shows a complete input schema, but get_ticket_price and calculate have schemas that cannot be verified in the provided source (they appear in airline.py which is not shown). Tool descriptions are present but minimal (10-35 chars), falling well below the 50-200 char baseline for LLM-optimized descriptions. No parameter descriptions are visible. The server mixes security analysis (semgrep) with unrelated airline functionality (ticket pricing, calculator), suggesting incomplete extraction or a composite demo server. Error handling is not documented. No output schemas are specified. Tool naming follows verb-noun convention (semgrep_scan, get_ticket_price, calculate), which is appropriate, but descriptions are too terse to guide LLM selection.
Evaluate a numeric expression - use this for example to make calculations about prices
Get the price of a ticket to a given city.
Scan code for security vulnerabilities using Semgrep static analysis
Tool descriptions are extremely terse (10-35 characters). LLM cannot determine when/why to select tools without richer context. Baseline is 50-200 chars; these fall far short.
No parameter descriptions are visible in source code. LLMs cannot infer that 'expr' for calculate means a numeric expression, or that 'config' for semgrep_scan must be exactly 'auto'. Parameter meaning must be explicit.
Only semgrep_scan has a visible input schema (in backend/mcp_servers.py). get_ticket_price and calculate are defined in airline.py which is not provided. Cannot verify schema presence, completeness, or correctness for 2 of 3 tools.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 36 | 2026-07-28+ | v2 |
No output schemas are documented. LLMs cannot plan downstream operations or extract result fields. semgrep_scan should document what the vulnerability report contains (title, severity, line number, etc.).
Server mixes unrelated domains: security analysis (semgrep_scan) + airline booking (get_ticket_price, calculate). This suggests incomplete extraction, demo code, or a composite test server. Production servers should have cohesive purpose.
semgrep_scan parameter 'config' requires exactly 'auto' but description says 'Semgrep configuration - must be exactly auto'. Should be an enum with single value 'auto' or a constant parameter, not a free-form string that invites hallucination.
semgrep_scan accepts nested objects (code_files array with filename, content, config). No example shown of what a valid request looks like or what the response structure is. LLMs will struggle to construct correct payloads.