updating-bostonaholic-skills

Warn

Audited by Gen Agent Trust Hub on Nov 15, 2026

Risk Level: MEDIUMREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The script scripts/update.mjs executes the command npx -y skills@1.8.1 remove ..., which downloads and runs code from the NPM registry at runtime.
  • [REMOTE_CODE_EXECUTION]: The skill clones a remote repository (github.com/bostonaholic/skills.git) and replaces local directories, including executable scripts in scripts/ and shared/ folders, across multiple platforms (Claude Code, Cursor, Codex, etc.). These updated scripts are subsequently executed by the agent.
  • [COMMAND_EXECUTION]: Uses the node:child_process module's spawnSync function to execute multiple CLI tools, including git, claude, codex, and npx, with arguments partially derived from remote data.
  • [EXTERNAL_DOWNLOADS]: Fetches release tags and full repository content from https://github.com/bostonaholic/skills.git to perform the update process.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted remote data which creates an attack surface for indirect prompt injection.
  • Ingestion points: Reads plugin.json and file contents from the cloned GitHub repository in scripts/update.mjs.
  • Boundary markers: Absent; the skill does not use delimiters or warnings when processing the remote skill instructions or manifests.
  • Capability inventory: Includes subprocess execution (spawnSync), arbitrary file writes/deletes (cpSync, rmSync, renameSync), and remote package execution (npx) in scripts/update.mjs.
  • Sanitization: Limited to basic regex validation for version tags and structure checks for the manifest file.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 15, 2026, 12:26 PM