awesome-adaptation-agentic-ai
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill downloads content from an external source via
git clone https://github.com/pat-jj/Awesome-Adaptation-of-Agentic-AI.gitto provide the agent with research paper data. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from an external repository which presents a surface for indirect prompt injection attacks. * Ingestion points: The
README.mdfile from the cloned GitHub repository (SKILL.md). * Boundary markers: Absent. Instructions suggest direct reading and searching of the downloaded content usingcatandgrepwithout delimiters or instructions to ignore embedded prompts. * Capability inventory: The skill utilizes shell commands for cloning and reading files, allowing the agent to ingest the file content. * Sanitization: Absent. No filtering or escaping is applied to the content of the downloaded README file before the agent processes it, meaning instructions hidden in the markdown could be executed by the agent.
Audit Metadata