skill-from-github
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The workflow involves ingesting and processing untrusted content from arbitrary GitHub repositories to inform the creation of new skills.
- Ingestion points: In Step 4, the agent is instructed to read README files, source code, examples, and dependency lists from external repositories identified via search.
- Boundary markers: The instructions do not specify the use of delimiters or "ignore embedded instructions" prompts when the agent processes the external project data.
- Capability inventory: The skill utilizes the
skill-creatortool to output new instructions based on the analysis of these untrusted sources. - Sanitization: There is no evidence of content filtering, escaping, or validation to prevent instructions hidden in the external code or documentation from influencing the agent's output.
Audit Metadata