The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides explicit instructions and command templates (grep, findstr, python -c) for the agent to execute via the exec tool to search local history files.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it retrieves and processes content from memory/HISTORY.md and memory/MEMORY.md, which store data from previous user interactions that could contain malicious instructions.
Ingestion points: memory/HISTORY.md and memory/MEMORY.md are read via read_file or exec in SKILL.md.
Boundary markers: The instructions do not specify the use of delimiters or warnings to ignore instructions embedded within the retrieved memory content.
Capability inventory: The agent is granted capabilities to execute shell commands via exec and perform file system modifications via write_file and edit_file across all files in the skill's scope.
Sanitization: There is no mention of sanitizing, escaping, or validating the content read back from the memory files before it is processed by the agent.

memory