byted-web-search
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXPOSURE]: The skill's execution script reads configuration from files such as
~/.openclaw/.envand/root/.openclaw/.envto retrieve API credentials. These credentials are used exclusively to authenticate with the vendor's official endpoints atmercury.volcengineapi.comandopen.feedcoopapi.com. - [INDIRECT_PROMPT_INJECTION]: Because the skill processes external web content, it presents an ingestion surface for potential indirect prompt injection. The skill instructions mitigate this by directing the agent to use multiple search strategies for cross-verification and to provide citations for all retrieved facts.
- [COMMAND_EXECUTION]: The skill uses a Python script to perform API operations. The script uses structured argument parsing to process user queries safely when invoked via the shell.
- [PERSISTENCE]: The instructions include a behavioral reinforcement mechanism that encourages the agent to save user search preferences to its long-term memory. This is a platform-supported feature for optimizing search workflows rather than a malicious persistence attempt.
Audit Metadata