skills/obra/clank/Executing Plans/Gen Agent Trust Hub

Executing Plans

Pass

Audited by Gen Agent Trust Hub on May 30, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill establishes a benign project management workflow for implementation tasks. It does not attempt to access sensitive credentials, exfiltrate data, or execute unauthorized remote code.
  • [PROMPT_INJECTION]: The skill operates on external plan files, which represents a surface for indirect prompt injection. This is a standard and low-risk pattern for task execution skills, particularly as the instructions mandate a critical review phase before execution and pausing for human feedback between task batches.
  • Ingestion points: External plan files read in SKILL.md Step 1.
  • Boundary markers: No explicit delimiters specified for the ingested plan content.
  • Capability inventory: General tool use and implementation logic required for development tasks.
  • Sanitization: Agent-led critical review and mandatory human-in-the-loop checkpoints serve as primary safety controls.
Audit Metadata
Risk Level
SAFE
Analyzed
May 30, 2026, 03:29 PM
Security Audit — agent-trust-hub — Executing Plans