The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest external repository instructions and reference documents (e.g., references/work-readiness.md). While this provides an attack surface for indirect prompt injection, the skill mitigates this risk through a multi-step workflow involving explicit manual approval gates, risk-gating logic, and a formal review process that specifically checks for security boundaries and blast radius.
[COMMAND_EXECUTION]: The skill uses tools like Bash to perform development tasks. These actions are restricted by a structured plan that requires user approval before branching or editing, ensuring all commands are within the scope of the intended development task.
[SAFE]: The skill references several external resources and routing identifiers such as pm:house-rules, harness:risk-gate, and harness:delegate. These are consistent with vendor-specific (studio-moser) workflow configurations and project management conventions, presenting no security risk.

dev-task