skills/g1joshi/agent-skills/awscli/Gen Agent Trust Hub

awscli

Pass

Audited by Gen Agent Trust Hub on Oct 11, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user to download the AWS CLI installer from awscli.amazonaws.com, which is an official and trusted domain for AWS tools.
  • [PRIVILEGE_ESCALATION]: The documentation includes the use of sudo to execute the installer command for the downloaded AWS CLI package. This is a standard requirement for system-level software installation on macOS and Linux.
  • [COMMAND_EXECUTION]: The skill utilizes shell commands like curl, aws, and jq to facilitate resource management and data processing, which is consistent with its purpose as a command-line utility guide.
  • [INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface by processing external data from AWS resource outputs (e.g., instance tags or metadata). While this data could be manipulated by an external actor to contain instructions, the skill uses structured querying (JMESPath) to mitigate direct execution risks.
  • Ingestion points: The skill ingests data via aws ec2 describe-instances and other CLI query commands in SKILL.md.
  • Boundary markers: No specific delimiters are used to wrap the output of AWS commands.
  • Capability inventory: Includes network access (AWS APIs, curl), file installation (sudo installer), and environment variable configuration (export).
  • Sanitization: The skill employs JMESPath projections and jq to extract specific fields, providing a layer of validation over raw text interpolation.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 11, 2026, 03:36 AM