dx-devops-work-item-manage

Pass

Audited by Gen Agent Trust Hub on Nov 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data originating from Salesforce DevOps Center work items.
  • Ingestion points: Work item subject and description fields are retrieved via sf devops work-item list and sf data query in SKILL.md (Phase 2, Step 3) and references/commit-api.md.
  • Capability inventory: The skill executes shell commands (sf, git, jq) and performs network operations via sf api request rest.
  • Boundary markers: The instructions specify wrapping variables in double quotes (e.g., --subject "<subject>") within shell commands, which acts as a basic delimiter but is not a robust security boundary against adversarial content.
  • Sanitization: No explicit sanitization or filtering logic is prescribed for data retrieved from work items before it is interpolated into subsequent command flags or displayed, creating a potential surface for indirect prompt injection if work item metadata contains malicious instructions or escape characters.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 14, 2026, 11:52 PM