The Agent Skills Directory

[CREDENTIALS_UNSAFE]: Hardcoded credentials found in configuration snippets for database high availability.\n
Evidence: password: strongpassword in SKILL.md (Patroni configuration snippet) and references/replica-topologies.md.\n- [COMMAND_EXECUTION]: Inclusion of administrative shell commands and utility calls for infrastructure management.\n
Evidence: `socat
/var/run/haproxy.sock <<< "set server be_app/app-1 state drain"inSKILL.md`.\n
Evidence: curl -X POST localhost:8080/drain in SKILL.md.\n
Evidence: pg_ctl promote -D /var/lib/postgresql/data in references/replica-topologies.md.\n- [PROMPT_INJECTION]: The skill processes user-supplied architectural requirements to generate configuration snippets without explicit sanitization or instructions for the agent to use boundary markers.\n
Ingestion points: User-provided architecture targets and parameters in SKILL.md (Agent Protocol section).\n
Boundary markers: Absent in the instructions for generating configuration and code output.\n
Capability inventory: Generation of shell commands (socat, curl), SQL queries (ALTER TABLE, UPDATE), and infrastructure configuration (HAProxy, Patroni) across all referenced files.\n
Sanitization: No instructions to validate or escape user inputs before interpolation into generated artifacts.

enterprise-high-availability