gait-session-tracking

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONNO_CODE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses shell commands to execute a Python script defined by the $GAIT_MCP_SCRIPT environment variable via an $MCP_CALL wrapper.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through the gait_record_turn tool.
  • Ingestion points: Data enters the system through the prompt and response fields in the gait_record_turn tool call (SKILL.md).
  • Boundary markers: There are no specific delimiters or instructions to ignore embedded commands within the recorded data.
  • Capability inventory: The skill possesses the capability to execute subprocesses (python3) and manage a local Git repository.
  • Sanitization: No explicit sanitization or filtering of the ingested turn data is performed before storage.
  • [NO_CODE]: The skill contains no executable code or scripts of its own, relying entirely on the host environment's configuration and pre-existing scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 12:31 AM