tao-list-capabilities
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to run local Python scripts (e.g., list_tao_capabilities.py) using bash. These commands rely on the TAO_SKILL_BANK_PATH environment variable, which determines the execution path. Evidence: Bash blocks in SKILL.md referencing scripts in ${TAO_SKILL_BANK_PATH:-~/tao-skill-bank}/scripts/.
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from script outputs and other skill files without using boundary markers or sanitization.
- Ingestion points: Outputs from scripts in the scripts/ directory and content from platform skill files (skills/platform/tao-run-on-*/SKILL.md).
- Boundary markers: Absent; no instructions are provided to delimit external data or warn the agent about embedded instructions.
- Capability inventory: The skill uses Bash for script execution and Read for file access.
- Sanitization: None; the agent is told to use the output as the 'source of truth' without validation.
Audit Metadata