skill-publish-notify
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python script
scripts/notify.pyto handle notification delivery via command-line arguments. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data via the
--textand--titleparameters, which creates a vulnerability surface if the inputs are generated by an untrusted or compromised agent in a multi-step chain. This could lead to deceptive notifications being sent to team channels. Evidence Chain: 1. Ingestion: Parameters inscripts/notify.py. 2. Boundary Markers: Absent. 3. Capability: Network POST requests to external IM platforms. 4. Sanitization: Usesjson.dumps()for payload construction. - [DATA_EXFILTRATION]: The skill is designed to transmit data (notifications) to external webhooks. It handles sensitive credentials such as Telegram bot tokens and webhook URLs. The documentation follows security best practices by advising users to store these in environment variables rather than hardcoding them.
Audit Metadata