pp-wavespeed
Warn
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to install a command-line interface (CLI) and an MCP server using remote package managers and source repositories.
- Instructions include
npx -y @mvanhorn/printing-press-library install wavespeed --cli-onlywhich fetches and executes code from the NPM registry. - Instructions include
go install github.com/mvanhorn/printing-press-library/...@latestwhich downloads and compiles source code directly from GitHub. - [DATA_EXFILTRATION]: The CLI tool documented in the skill supports a
--deliver <sink>flag that routes command output to external destinations. - The
webhook:<url>sink allows the agent to POST the output body (which may contain sensitive API results or generation data) to any provided URL. - The
file:<path>sink allows the agent to write command output to any local file path. This could be used to write data to shared directories or attempt to overwrite configuration files. - [COMMAND_EXECUTION]: The skill's primary function is to execute the
wavespeed-pp-clibinary. It provides a natural language interface (whichcommand) and an agent mode (--agent) specifically designed to facilitate autonomous execution by the AI agent. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data and provides powerful capabilities, creating a surface for indirect prompt injection attacks.
- Ingestion points: The skill ingests natural language prompts, briefs, and search queries via commands like
plan brief-to-shotlist --prompt,library search, andwhich. - Boundary markers: The provided command templates do not include explicit instructions or delimiters to ignore embedded commands within the processed data.
- Capability inventory: The skill has access to execute the CLI binary, write to the local file system (
--deliver file:), and perform network POST requests (--deliver webhook:). - Sanitization: The skill documentation does not mention sanitization or validation of the input strings before they are processed by the CLI tool or downstream models.
Audit Metadata