The Agent Skills Directory

Indirect Prompt Injection (HIGH): The skill is vulnerable to instructional takeover via processed test data.
Ingestion points: Untrusted data enters the context through testany_get_execution and testany_get_execution_case calls as described in the workflow (SKILL.md).
Boundary markers: Absent. There are no instructions to the agent to treat test names, logs, or error messages as data rather than instructions.
Capability inventory: The agent can trigger external pipeline executions (testany_execute_pipeline) and provide recommendations based on test failures (SKILL.md).
Sanitization: None specified. Maliciously crafted test case names or failure logs could contain instructions that redirect the agent's workflow.
Command Execution (SAFE): The skill uses predefined tool calls for interaction and does not appear to spawn arbitrary shell processes.
External Downloads (SAFE): No remote scripts or unverified package installations are referenced.

tests