split-task

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill establishes a surface for processing external project data and task descriptions.\n
  • Ingestion points: User-provided 'Outcome' and 'Ticket / PR' reference data captured at runtime.\n
  • Boundary markers: The skill does not define specific delimiters or instructions to ignore commands within the ingested text.\n
  • Capability inventory: The skill triggers downstream tool usage (/design, /implement) and can write artifacts to the filesystem if requested.\n
  • Sanitization: Input data is processed directly as part of the task decomposition without explicit filtering.\n- [NO_CODE]: The skill consists entirely of instructional markdown and documentation templates. It does not contain any executable scripts, binaries, or source code.\n- [SAFE]: The skill configuration includes disable-model-invocation: true, which effectively scopes and restricts the agent's behavior for this specific task.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 04:39 PM
Security Audit — agent-trust-hub — split-task