The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes experiment metadata (names, descriptions) from the experiment-list tool to identify IDs.
Ingestion points: Data returned by the experiment-list tool as described in SKILL.md.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are present in the skill definition.
Capability inventory: The skill uses experiment-list, experiment-get-by-flag-key, and experiment-get tools; it also references potentially destructive actions like 'delete', 'ship', or 'end' in linked workflows.
Sanitization: There is no evidence of sanitization or filtering applied to the experiment data before the agent interprets it.

finding-experiments