The Agent Skills Directory

Data Exposure & Exfiltration (LOW): Hardcoded default credentials for local services (PostgreSQL, MinIO, and Grafana) are provided in the documentation for development purposes. Evidence found in rules/service-architecture.md includes credentials like polar/polar and polar/polarpolar.
Indirect Prompt Injection (LOW): The skill exhibits a surface for indirect prompt injection by reading untrusted container logs. * Ingestion points: dev docker logs (referenced in SKILL.md, rules/view-logs.md). * Boundary markers: None identified in command instructions. * Capability inventory: Arbitrary shell execution (dev docker shell), service lifecycle management, and environment destruction (dev docker cleanup). * Sanitization: No sanitization or validation of log content is specified before processing.
Dynamic Execution (LOW): The skill explicitly allows for the execution of arbitrary commands within containerized environments. Evidence in SKILL.md and rules/troubleshooting.md includes the dev docker shell {service} command.

local-environment