The Agent Skills Directory

[COMMAND_EXECUTION]: The skill directs the agent to utilize shell commands for system investigation and safety procedures, specifically using cp for file backups, lsof for checking file locks, and grep for dependency analysis.- [PROMPT_INJECTION]: The skill employs strong imperative language and Iron Rules to override default agent behaviors and enforce its own operational framework. It also creates a surface for indirect prompt injection. 1. Ingestion points: Error messages, source code, and web search results (referenced in SKILL.md). 2. Boundary markers: Absent. 3. Capability inventory: Bash (cp, grep, lsof, cat), curl, and WebSearch (SKILL.md). 4. Sanitization: Absent.- [EXTERNAL_DOWNLOADS]: The instructions mandate the use of WebSearch and curl to retrieve external data for error verification and documentation, introducing unverified remote content into the agent context.- [DATA_EXFILTRATION]: The skill encourages the use of curl and WebSearch to interact with external services for verification purposes. If combined with instructions to read sensitive local files like configuration files or environment variables, this creates a potential pathway for data exposure.

yes