The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute grep commands for parsing project documentation (.claude/docs/effects-map.md) to dynamically identify reserved settings.
[REMOTE_CODE_EXECUTION]: Sources an external shell script (.claude/hooks/yaml-helper.sh) to handle YAML parsing and logic. While this executes code outside the primary skill file, it is a project-local dependency intended for CCGS framework operations.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external configuration files that could potentially contain malicious instructions.
Ingestion points: Reads project.yaml and project.local.yaml via the Read tool and helper functions.
Boundary markers: Not explicitly defined in the prompt interpolation, but settings are validated against allowed enums.
Capability inventory: Includes Bash command execution and file modification via Write and Edit tools.
Sanitization: Employs schema validation (validate_enum_value) and mandatory user confirmation via AskUserQuestion for all write operations to mitigate unauthorized changes.

settings