The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to ingest and process log data from external sources. Malicious actors could inject instructions into logs that the agent might inadvertently follow during analysis or automated remediation.
Ingestion points: Processes application, infrastructure, and audit logs (referenced in SKILL.md and references/README.md).
Boundary markers: Absent. There are no instructions for the agent to treat log content as untrusted data or use specific delimiters.
Capability inventory: Executes shell scripts (scripts/create-dashboard.sh), calls npm scripts for metrics extraction and correlation, and describes automated remediation actions like restarting services or scaling infrastructure.
Sanitization: Absent. No evidence of escaping or validating log content before it influences agent decisions.
[COMMAND_EXECUTION]: The skill contains a shell script scripts/create-dashboard.sh that generates JSON and YAML configuration files based on user-provided arguments. It also documents various npm run commands that execute project-specific tooling for metrics extraction and incident management.

observability-logging