quark-download
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions require the agent to execute a local Python script (
scripts/quark_search.py) with various subcommands to perform searches, validate links, and trigger saves. The parameters for these commands, such as search keywords and share IDs, are derived from user input or external API responses. - [INDIRECT_PROMPT_INJECTION]: The skill acts as a bridge to external, untrusted search results from the PanSou aggregation API (
s.panhunt.com) and file metadata from the Quark share service (drive-pc.quark.cn). These results, including user-generated fields like 'note' and 'file_name', are presented directly to the AI agent. - Ingestion points: Data enters the agent's context through the output of
cmd_searchandcmd_detailinscripts/quark_search.py. - Boundary markers: No specific delimiters or instructions to ignore embedded prompts are defined in the skill for these data ingestion points.
- Capability inventory: The agent can execute the included Python script which performs network operations via
urlliband manages a local cache file at~/.cache/quark-search/health.json. - Sanitization: There is no evidence of sanitization or filtering of the text returned from external APIs before it is processed by the AI.
Audit Metadata