agent-academy-stargazers
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONPRIVILEGE_ESCALATIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- System Package Installation & Command Execution: The skill instructs the agent to install the GitHub CLI (
gh) using system package managers (e.g.,brew) if it is not present. This involves executing shell commands and may require elevated privileges depending on the user's environment. - External Service Interaction: The skill fetches data from the GitHub API using the
ghCLI. While this involves network operations, it targets a well-known service to retrieve information related to the repository specified in the skill's description. - Indirect Prompt Injection Surface: The skill ingests and processes data from external GitHub profiles (stargazer lists). Although the skill uses specific filters to extract only certain fields like usernames and URLs, processing data from external sources is a common consideration for agent safety.
- Ingestion points: Data is fetched via
gh apiand stored intemp/stargazers.json(SKILL.md). - Boundary markers: The skill uses a JQ filter to strictly define the expected data structure (
login,type,url). - Capability inventory: The skill uses shell execution for
ghcommands and package management (SKILL.md). - Sanitization: The skill employs specific field selection via JQ to limit the data processed by the agent.
Audit Metadata