The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to scan and ingest raw patterns from an existing (brownfield) codebase to generate Context Directive Records (CDRs). This creates an attack surface where malicious instructions embedded in the code being analyzed could influence the synthesis agent's output.
Ingestion points: Phase 5 (Discovery Agent) and Phase 6 (Pattern Agent) involve scanning the codebase for reusable rules, personas, and examples.
Boundary markers: The instructions do not specify any boundary markers (like delimiters or 'ignore' instructions) to isolate the agent from instructions potentially hidden within the code comments or documentation it processes.
Capability inventory: The skill has the ability to execute shell scripts (Phase 1) and write files to the local file system (Phase 9).
Sanitization: There is no evidence of sanitization or filtering applied to the external codebase content before it is processed by the synthesized multi-agent pipeline.
[COMMAND_EXECUTION]: The skill executes platform-specific shell scripts (scripts/bash/setup-levelup-init.sh or scripts/powershell/setup-levelup-init.ps1) during its initialization phase to resolve project roots, directives paths, and environment variables.

levelup-init