dex-plan
Warn
Audited by Gen Agent Trust Hub on Nov 13, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to use
npx @zeeg/dexif the localdexcommand is unavailable. This downloads and executes a package from the public npm registry under an organization scope (@zeeg) that is not identified as a trusted vendor or associated with the author's known resources. - [REMOTE_CODE_EXECUTION]: The use of
npxto run a package from an unverified source constitutes remote code execution, as the package content is fetched and executed at runtime. - [INDIRECT_PROMPT_INJECTION]: The skill reads external markdown planning documents and extracts text to populate the name and description arguments of shell commands, creating a vulnerability surface. * Ingestion points: Markdown files from paths like
~/.claude/plans/*.mdor user-specified documents likeSPEC.md. * Boundary markers: Absent; the agent is directed to read raw markdown content without delimiters or instructions to ignore embedded commands. * Capability inventory: The agent executes shell commands (dex create,dex plan) using strings extracted directly from the markdown headings and lists. * Sanitization: Absent; there are no instructions to escape shell metacharacters (such as double quotes or backticks) or sanitize the extracted text before passing it to the CLI tool.
Audit Metadata