skills/webpro/skills/eval-skill/Gen Agent Trust Hub

eval-skill

Warn

Audited by Gen Agent Trust Hub on Nov 7, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONCREDENTIALS_UNSAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The evaluation script scripts/eval-trigger executes arbitrary shell commands defined in the setup field of evaluation case files. While this requires the --allow-setup flag, it allows the skill to run code with host environment privileges.
  • [CREDENTIALS_UNSAFE]: The script scripts/eval-trigger contains logic to access sensitive platform credentials. It can read ~/.claude/.credentials.json and invoke the macOS security command to retrieve passwords from the login keychain. This is used for the --auth-passthrough feature to authenticate agents in isolated environments.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted datasets from trigger.json to evaluate agent triggers. This represents a vulnerability surface where external data is directly interpolated into agent prompts.
  • Ingestion points: trigger.json (specifically the query and files fields).
  • Boundary markers: Absent; the query is passed directly to the agent CLI as a prompt.
  • Capability inventory: The harness can execute shell commands (subprocess.run), write to the filesystem, and facilitate network access via the agent CLI.
  • Sanitization: No explicit prompt sanitization is performed on the input query before evaluation.
  • [DYNAMIC_EXECUTION]: The setup phase in scripts/eval-trigger employs subprocess.run with shell=True to execute commands provided as strings in external JSON files, facilitating the dynamic execution of potentially untrusted code.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 7, 2026, 09:23 AM