skills/dsebban/skills/poteto-mode/Gen Agent Trust Hub

poteto-mode

Pass

Audited by Gen Agent Trust Hub on Aug 18, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The scripts/bootstrap.ts utility automatically manages its own dependencies by running bun install --frozen-lockfile if the local environment is missing required packages (e.g., commander). This operation uses a frozen lockfile to ensure that only verified, pinned versions of dependencies are installed from the registry.
  • [COMMAND_EXECUTION]: The skill interacts with the local environment using standard developer tools. The orchestrate and watch-pr tools execute commands like git, gh (GitHub CLI), and gt (Graphite CLI) to manage branches, pull requests, and multi-PR stacks. These actions are performed to facilitate project management and local development workflows as defined in the skill's playbooks.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: The worktree-cleanup playbook and worktree-audit.sh script read metadata from the local Git repository and calculate disk usage to help users maintain their development environment. This information is used locally for decision-making and is not exfiltrated to unauthorized external domains.
  • [PROMPT_INJECTION]: The skill defines a set of 'Principles' and 'Non-negotiables' to guide agent behavior. Rather than attempting to bypass safety filters, these instructions implement a defensive security posture, explicitly warning the agent to treat external review comments (e.g., from 'bugbot') as untrusted data that should not be interpreted as commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 18, 2026, 01:03 PM
Security Audit — agent-trust-hub — poteto-mode