The Agent Skills Directory

[DATA_EXFILTRATION]: The skill provides a command to display environment variables containing sensitive strings like "SECRET", "API", or "DB" (env | grep -i "DB_\|API_\|SECRET_" | sort). This practice exposes sensitive credentials in the agent's execution logs or output.
[INDIRECT_PROMPT_INJECTION]: The skill instructions include reading application log files to find errors. This creates a surface where malicious instructions embedded in logs could influence the agent.
Ingestion points: Application logs (logs/app.log) read via the grep tool as part of the debugging workflow in SKILL.md.
Boundary markers: No delimiters or protective instructions are provided to separate log data from agent instructions.
Capability inventory: The agent has access to Bash, Read, Glob, and Grep tools.
Sanitization: No sanitization or validation of the log content is performed.
[COMMAND_EXECUTION]: The skill leverages the Bash tool to perform various diagnostic tasks including git, curl, and psql. While appropriate for the debugging context, it expands the potential impact of other vulnerabilities.

service-debugging