file-operations
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill enables the agent to read and process local file content, which introduces a risk of indirect prompt injection. If an analyzed file contains malicious instructions, the agent may inadvertently follow them.\n
- Ingestion points: Data returned from
ReadandGrepoperations on local files.\n - Boundary markers: No delimiters or instructions are provided to distinguish file data from system prompts.\n
- Capability inventory: File system access via shell commands (
stat,wc,ls,du,find) and native tools (Read,Grep,Glob).\n - Sanitization: No input validation or instruction-filtering is applied to the file contents.\n- [COMMAND_EXECUTION]: The skill provides structured bash command patterns for the agent to execute, such as
find,stat, andwc, which directly interact with the filesystem to gather metadata.
Audit Metadata