The Agent Skills Directory

[PROMPT_INJECTION]: The SKILL.md file contains forceful directives intended to override the agent's operational logic and reasoning. It uses phrases like "⚠️ MANDATORY SKILL INVOCATION ⚠️", "YOU MUST invoke this skill (NOT optional)", and claims that failure to do so "violates your operational requirements." These instructions attempt to coerce the agent into specific tool usage regardless of context or safety considerations.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external torrent and usenet indexers via search results.
Ingestion points: Search results (titles, descriptions, info links) returned by scripts/prowlarr-api.sh and displayed to the agent.
Boundary markers: Absent. The skill does not provide delimiters or instructions to the agent to ignore potential commands embedded in release metadata.
Capability inventory: The skill provides administrative capabilities including deleting indexers, disabling indexers, and syncing applications, which could be abused if an injection is successful.
Sanitization: The script uses jq to parse JSON results and encode query parameters, which prevents structural injection but does not mitigate natural language instruction injection.

prowlarr