Engineering Evidence Database
Pass
Audited by Gen Agent Trust Hub on Aug 23, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill defines a process for the agent to ingest project data from external third-party sources (e.g., GitHub, Notion, Confluence, JIRA). This represents an attack surface for indirect prompt injection, as malicious instructions within these external sources could influence agent behavior. Ingestion points: The
evidence_sourcesobject inSKILL.mdlists various external platform URLs. Boundary markers: The instructions lack explicit delimiters or "ignore embedded instructions" warnings for processing external data. Capability inventory: The skill itself lacks executable code but is designed for an agent environment capable of fetching and reading web content and files. Sanitization: No sanitization or validation rules are defined for the ingested content. - [NO_CODE]: The skill consists solely of instructional Markdown and JSON templates, containing no executable scripts, binaries, or shell commands.
Audit Metadata