The Agent Skills Directory

[COMMAND_EXECUTION]: The utility script scripts/generate_test_suite.py contains a generate_script method that assembles shell scripts from string templates and explicitly applies executable permissions using chmod 0o755. This creates a mechanism where the agent can generate and immediately execute arbitrary shell logic based on its instructions or processed data.
Evidence: scripts/generate_test_suite.py lines 78-98.
[PROMPT_INJECTION]: The skill is designed to ingest and parse untrusted external data, specifically OpenAPI/Swagger specifications and GraphQL schemas, to automate test case generation. This creates a surface for indirect prompt injection where malicious instructions embedded in the 'description' or other metadata fields of a specification could attempt to influence the agent's actions or output.
Ingestion points: The agent is instructed to parse OpenAPI specs and introspect GraphQL schemas (SKILL.md, Instructions step 1).
Boundary markers: Absent. There are no instructions for the agent to use delimiters or ignore instructions found within the processed specifications.
Capability inventory: The skill has access to Bash, Write, and Edit tools, which could be leveraged if an injection is successful.
Sanitization: Absent. The provided logic does not include validation or sanitization of specification content before processing.

automating-api-testing