The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection because it automates file modifications based on findings derived from external target skills. \n
Ingestion points: Findings and evidence confirmed by the adaptogen-critique process. \n
Boundary markers: The instructions do not specify any delimiters or instructions to disregard instructions embedded within the findings. \n
Capability inventory: The skill possesses powerful file modification (Write, Edit) and execution (Bash) tools. \n
Sanitization: There is no logic provided to sanitize or validate the findings before they are used to update target files. \n- [COMMAND_EXECUTION]: The skill utilizes the system shell for environment inspection and script generation. \n
It employs the Bash tool with git subcommands to evaluate task determinism through history analysis. \n
It is explicitly instructed to generate and write new executable scripts into a target skill's scripts/ directory.

adaptogen-land