The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of multiple tools and plugins from unknown GitHub repositories and NPM packages. Examples include github.com/cursor/plugins, Prinova/pi-agent-codebase-workflows, and NPM packages like @earendil-works/pi-agent-browser and @tintinweb/pi-tasks. These sources are not recognized as trusted vendors or well-known services.
[COMMAND_EXECUTION]: The skill uses shell commands for its core logic, including bash, grep, and node -e. For example, it uses node -e to parse the stelow.json configuration file to resolve workflow directories and review modes. It also coordinates multiple subagents which can execute arbitrary shell tasks.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted product specifications (spec-product.md) to perform gap analysis. This content is used to generate reports and is eventually merged back into the plan document.
Ingestion points: The skill reads spec-product.md in critique:20 and passes it to parallel subagents in critique:30 for analysis.
Boundary markers: No specific boundary markers or 'ignore' instructions are provided when processing the input files.
Capability inventory: The skill can execute arbitrary shell commands, launch subagents, write to the filesystem, and interact with websites via agent_browser.
Sanitization: The instructions do not include any steps to sanitize or validate the content of the spec-product.md file before processing or merging it.

stelow-product-plan-critique