kagi-account-config
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from Kagi account settings (lenses, custom bangs, and assistant configurations) which could theoretically contain malicious instructions if the account is compromised or malicious data is synchronized.
- Ingestion points: The
kagi lens list,kagi assistant custom list,kagi site-pref list, andkagi bang custom listcommands inSKILL.mdingest external data from the Kagi API into the agent context. - Boundary markers: The instructions do not specify boundary markers or delimiters for the output of these commands.
- Capability inventory: The agent has access to the
kagiCLI viaBash(kagi:*), allowing it to read and write account configurations. - Sanitization: No sanitization or validation of the API responses is mentioned before processing.
- [COMMAND_EXECUTION]: The skill uses the
kagiCLI tool for all operations. This is the intended primary purpose and the instructions include best practices, such as warning the agent never to print stored credentials or pass secrets on shared command lines.
Audit Metadata