agently-knowledge-base-and-rag

Pass

Audited by Gen Agent Trust Hub on Mar 11, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSNO_CODE
Full Analysis
  • [PROMPT_INJECTION]: The documentation describes a standard Retrieval-Augmented Generation (RAG) workflow in references/retrieval-to-answer.md where retrieved data is injected into the prompt. This pattern inherently possesses a surface for indirect prompt injection. * Ingestion points: Data enters the agent context through the info() method. * Boundary markers: The prompt pattern uses simple interpolation ({retrieval_results}) without robust delimiters. * Capability inventory: The skill describes retrieval and answering; no direct code execution or file-write capabilities are defined. * Sanitization: The documentation does not detail specific sanitization or validation of retrieved content.
  • [EXTERNAL_DOWNLOADS]: The skill references official vendor resources, including the agently.tech documentation site and the AgentEra/Agently GitHub repository. These are verified vendor-owned assets and do not involve untrusted third-party sites.
  • [NO_CODE]: The skill consists entirely of markdown files providing instructions and references, with no executable scripts, binary files, or hidden code detected.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 11, 2026, 05:52 PM