The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes OpenAPI specifications (JSON files) fetched from MikoPBX instances, which provides an ingestion point for untrusted data. Maliciously crafted specifications could theoretically embed instructions in fields like summary, description, or example to influence agent behavior during analysis.
Ingestion points: The MikoPBXOpenAPIAnalyzer class in scripts/openapi_analyzer.py loads and parses the /tmp/mikopbx_openapi.json file.
Boundary markers: No explicit delimiters or boundary markers are used when interpolating specification content into analysis reports or generated scripts.
Capability inventory: The skill has access to Bash via allowed-tools and demonstrates file-writing capabilities in its integration examples.
Sanitization: The skill relies on standard JSON parsing; however, text content from the specification is used directly in generated Python test code and output summaries with minimal transformation.
[DYNAMIC_EXECUTION]: The skill includes patterns for generating and executing code based on external specification data, primarily for the purpose of API test generation and code-to-spec synchronization.
Evidence: In examples/integration-examples.md, the skill demonstrates a workflow that generates a complete Python test suite (tests/api/test_extensions_openapi.py) by interpolating values from the OpenAPI specification into a script template. It also provides examples of executing inline PHP code blocks within a Docker container using docker exec.

openapi-analyzer