agentforce-skills-research
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from multiple sources which could contain malicious instructions designed to influence the agent's behavior.
- Ingestion points: The skill retrieves data from Salesforce Help, Trailhead, internal git repositories (git.soma), Slack channels, and Google Docs (SKILL.md).
- Boundary markers: The instructions do not define clear delimiters or "ignore instructions" warnings when processing the retrieved research materials.
- Capability inventory: The skill has access to powerful tools including
Write,Edit,WebFetch, andAgent, which could be exploited if the agent executes instructions embedded in search results. - Sanitization: No explicit sanitization, filtering, or validation of the content retrieved from external or internal sources is implemented before the agent processes it or saves it to research artifacts.
Audit Metadata