competition-web-runtime
Pass
Audited by Gen Agent Trust Hub on Jun 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides Python and shell code snippets in 'references/cookie-hmac-key-reuse-auth-bypass.md' intended for the agent to forge authentication cookies and interact with web APIs during a challenge. These are legitimate within the skill's stated CTF purpose.
- [PROMPT_INJECTION]: The skill facilitates processing untrusted data from web applications, creating a surface for indirect prompt injection. \n
- Ingestion points: Browser storage (LocalStorage, IndexedDB), HTTP headers, HTML content, and server responses as specified in 'SKILL.md' and 'references/routing-runtime.md'. \n
- Boundary markers: The instructions do not define delimiters to separate untrusted web data from agent instructions. \n
- Capability inventory: The skill uses tools like 'curl' and Python for web interaction and data processing. \n
- Sanitization: No specific sanitization or validation of the ingested external content is mentioned.
Audit Metadata