axiom-xcode-mcp-tools
Warn
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: MEDIUMDYNAMIC_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill documents the
ExecuteSnippettool, which allows the execution of Swift code snippets in a REPL environment. While the skill notes the environment is sandboxed, this provides a pathway for dynamic code execution. - [COMMAND_EXECUTION]: The skill includes instructions for using
XcodeRM, a tool capable of performing destructive file and directory deletions within the project. - [EXTERNAL_DOWNLOADS]: The documentation references an external GitHub repository (
github.com/SoundBlaster/XcodeMCPWrapper) as a recommended proxy, which is not identified as a trusted vendor. - [INDIRECT_PROMPT_INJECTION]: The automated BuildFix and TestFix workflows ingest untrusted data (build logs, test failures) to drive code modifications, establishing an indirect prompt injection surface.
- Ingestion points:
XcodeRead,GetBuildLog,XcodeListNavigatorIssues, andDocumentationSearchinSKILL.md. - Boundary markers: No boundary markers or instructions to ignore embedded commands are specified.
- Capability inventory:
XcodeWrite,XcodeUpdate,XcodeRM,XcodeMV,ExecuteSnippet, andBuildProjecttools are available inSKILL.md. - Sanitization: The skill does not provide instructions for sanitizing or escaping the data ingested from logs or external documentation before processing.
Audit Metadata