The Agent Skills Directory

[SAFE]: The skill serves as a documentation guide for the Eino framework by CloudWeGo. No malicious instructions, obfuscation, or persistence mechanisms were found.
[EXTERNAL_DOWNLOADS]: The skill references package installation commands (e.g., go get github.com/cloudwego/eino@latest) in reference/quick-start.md. These downloads target the official repositories of the framework developer and are standard for software development.
[INDIRECT_PROMPT_INJECTION]: The skill defines components like Retriever and Document Loader in SKILL.md and reference/schema.md that are designed to ingest untrusted external data (e.g., documents from the web or filesystems).
Ingestion points: schema.Document content and schema.UserMessage inputs are processed by components.
Boundary markers: The examples show the use of ChatTemplate with Jinja2-style variables, which can help delimit input, though specific safety instructions are not included in the snippets.
Capability inventory: The framework supports tool execution through InvokableTool and ChatModelAgent as described in SKILL.md.
Sanitization: Standard component interfaces are defined, though the specific examples do not demonstrate input sanitization or validation routines.
[CREDENTIALS_UNSAFE]: The code examples in reference/quick-start.md use the placeholder "your-api-key" for API configuration. This is a standard documentation practice and does not expose actual sensitive credentials.

eino-guide