The Agent Skills Directory

EXTERNAL_DOWNLOADS (HIGH): The skill requires the installation of multiple packages from the @thesysai and @crayonai scopes, and clones a template from github.com/thesysdev. These sources are not included in the global trusted scope.
COMMAND_EXECUTION (HIGH): Setup instructions include npx create-c1-app, npm install, and pip install -r requirements.txt. These commands execute scripts from unverified third-party sources at runtime.
PROMPT_INJECTION (HIGH): (Category 8
Indirect) The skill's core function is to build a Generative UI pipeline, creating a significant attack surface for indirect prompt injection.
Ingestion points: User-provided prompts and thread data are ingested via req.json() in app/api/chat/route.ts and passed to a third-party API.
Boundary markers: Absent. The provided logic does not demonstrate the use of delimiters or instructions to ignore embedded commands within the data being visualized.
Capability inventory: The <C1Component> handles onAction (UI events) and updateMessage, which is explicitly documented to 'Persist state changes to database'. This gives AI-generated content direct 'write' capabilities to the backend.
Sanitization: There is no evidence of sanitization or validation of the DSL/XML structure returned by the C1 API before it is rendered into interactive React components.
CREDENTIALS_UNSAFE (LOW): While the skill uses appropriate placeholders like <your-api-key>, it encourages storing sensitive API keys in environment variables via shell exports (export THESYS_API_KEY), which may be logged in shell history.

thesys-c1-genui