jmb-setup

Pass

Audited by Gen Agent Trust Hub on Nov 21, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the gh CLI and git to perform project exploration and administrative tasks. Specifically, it uses gh api to modify repository-level merge settings, such as enforcing squash-merges and branch deletion. These actions are clearly documented and require human confirmation before execution.\n- [EXTERNAL_DOWNLOADS]: The instructions involve running npx skills update and npx skills add, which download and execute tools from the npm registry to manage the skill kit.\n- [PERSISTENCE]: The skill installs a PreToolUse hook in .claude/settings.json. This hook points to block-dangerous-git.sh, a script designed to intercept and block destructive or unauthorized Git commands. This configuration ensures the security guardrail remains active across sessions.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes data from the local project environment to populate its configuration templates, creating an attack surface for indirect injection.\n
  • Ingestion points: The skill reads local project files (package.json, documentation, etc.) and repository metadata via shell commands to determine the current state and fill in template slots.\n
  • Boundary markers: Generated files use specific <!-- jmb-setup:start --> and <!-- jmb-setup:end --> markers to identify and scope automated edits.\n
  • Capability inventory: The skill can execute shell commands, perform GitHub API calls, and write to various configuration and documentation files.\n
  • Sanitization: All changes are presented to the user for review and explicit confirmation before any files are modified or API calls are made.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 21, 2026, 06:51 PM