pre-commit-smoke-test

Warn

Audited by Gen Agent Trust Hub on Jun 22, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses sudo to perform several high-privilege system operations:
  • Stopping the existing observe-agent service via systemctl to prevent port conflicts.
  • Running the local agent binary with root privileges to enable access to system log directories and bind to privileged ports.
  • Creating and managing test log files in the /var/log/ directory tree.
  • Terminating running agent processes using kill and pgrep.
  • [COMMAND_EXECUTION]: The skill triggers the compilation of the agent binary from source code using the make build command.
  • [PROMPT_INJECTION]: The skill is subject to indirect prompt injection risks because it reads and parses data from system logs (/var/log/**/*.log) to verify status messages.
  • Ingestion points: The skill reads from standard system logs and a generated test log file at /var/log/smoke-test/smoke-test.log.
  • Boundary markers: No specific delimiters or markers are defined to isolate log content from instructions.
  • Capability inventory: The skill possesses the ability to execute shell commands with sudo, compile code, and delete system files.
  • Sanitization: There is no evidence of sanitization or filtering performed on the log data before it is processed.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jun 22, 2026, 07:57 PM
Security Audit — agent-trust-hub — pre-commit-smoke-test