Community extension of CrowdStrike's falcon-mcp MCP server with full Falcon API coverage (~1,296 tools)
Server has 8 tools with mixed quality. Tool naming is inconsistent: generic names like 'falcon_check_connectivity' lack verb clarity, while auto-generated names like 'download_export_file_mixin0' and 'launch_export_job_mixin0' are opaque and don't follow verb_noun conventions. Descriptions vary widely, some are detailed (e.g., search_serverless_vulnerabilities at ~280 chars), others are minimal (falcon_check_connectivity, falcon_list_modules at ~50 chars). Input schemas are present but sparse: most tools accept only 1-3 parameters with minimal type info. The 'body' parameter in launch_export_job_mixin0 is a bare dict with no field-level schema. Output schemas are not documented anywhere. Error handling is absent, no recovery guidance, no categorization of retryable vs fatal errors. Tool composition is weak: no clear chaining between discovery tools (list_modules, list_enabled_modules) and action tools. Auto-generated module names (mixin0 suffix) suggest code generation without human review for LLM usability.
Download an export file
Check connectivity to the Falcon API.
Lists enabled modules in the falcon-mcp server. These modules are determined by the --modules flag when starting the server. If no modules are specified, all available modules are enabled.
Lists all available modules in the falcon-mcp server.
Launch an export job of a Lambda Security resource. Maximum of 1 job in progress per resource. Use expand_vulnerabilities=true to get detailed vulnerability information.
Query export jobs entities
Auto-generated tool names (mixin0 suffix) are opaque and violate verb_noun convention. Names like 'download_export_file_mixin0' and 'launch_export_job_mixin0' do not clearly convey action or resource. LLMs cannot infer intent from 'mixin0' suffix.
Output schemas are not documented. Tools return 'list[dict] | dict' with no field-level schema. LLMs cannot plan downstream calls or extract required IDs without knowing response structure.
Minimal descriptions for discovery tools. 'falcon_check_connectivity' (20 chars), 'falcon_list_modules' (30 chars) lack context for when/why to call them. Baseline is 194 chars; these are 85% below average.
Inferred effective spec: 2026-07-28+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 47 | 2026-07-28+ | v2 |
Read export jobs entities
Search for vulnerabilities in your serverless functions across all cloud service providers. This endpoint provides security information in SARIF format, including: - CVE IDs for identified vulnerabilities - Severity levels - Vulnerability descriptions - Additional relevant details IMPORTANT: You must use the `falcon://serverless/vulnerabilities/fql-guide` resource when you need to use the `filter` parameter. This resource contains the guide on how to build the FQL `filter` parameter for the `falcon_search_serverless_vulnerabilities` tool.
No error handling or recovery guidance. Tools return raw API responses with no actionable error messages. If a query fails, LLM has no guidance on retry strategy or alternative approaches.
Body parameter in launch_export_job_mixin0 is a bare dict with no field-level schema. Description says 'per the CrowdStrike API schema' but does not specify which fields are required, their types, or valid values. LLM must guess or hallucinate the structure.