workflow-yylo

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes workflow records fetched from an external source and user input, creating a potential surface for indirect prompt injection.\n
  • Ingestion points: Data retrieved via yy ledger workflow search or get in SKILL.md, and user-provided input passed through the $ARGUMENTS placeholder.\n
  • Boundary markers: The skill instructions include use of the --validated flag and explicit warnings against storing executable shell as a substitute for validated data.\n
  • Capability inventory: Uses the yy ledger CLI for searching, retrieving, and creating workflow records.\n
  • Sanitization: Enforces strict YAML validation, rejecting unsafe structures like aliases or anchors, and utilizes schema-based normalization (schema_version: v1).\n- [COMMAND_EXECUTION]: The skill employs vendor-specific CLI tools (yy ledger, yylo-ledger) to interact with the ledger system. These commands are necessary for the skill's primary function of workflow record management.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 11:12 PM
Security Audit — agent-trust-hub — workflow-yylo