Skills
Navigate
skills/brycewang-stanford/auto-empirical-research-skills/macro-briefing/Gen Agent Trust Hub

macro-briefing

Fail

Audited by Gen Agent Trust Hub on Jun 22, 2026

Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches parameter data and updates from a GitHub repository.
  • Evidence: git clone https://github.com/charlescoverdale/econstack-data.git and instructions to git pull updates for the econstack tool.
  • [COMMAND_EXECUTION]: Executes several local binaries silently during the skill's preamble.
  • Evidence: Runs ~/.claude/skills/econstack/bin/econstack-update-check and ~/.claude/skills/econstack/bin/econstack-learnings-read without user interaction.
  • [COMMAND_EXECUTION]: Uses dynamic shell execution to load environment configurations.
  • Evidence: eval "$(~/.claude/skills/econstack/bin/econstack-slug)" executes the output of a local binary as shell code.
  • [REMOTE_CODE_EXECUTION]: Implements a self-update and execute pattern from an unverified source.
  • Evidence: The skill instructs the agent to check for updates and prompts the user to pull code from a remote repository, then immediately executes binaries and eval statements from that updated directory, allowing for the execution of arbitrary remote content.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Jun 22, 2026, 03:11 AM
Security Audit — agent-trust-hub — macro-briefing