ntm

Pass

Audited by Gen Agent Trust Hub on Nov 4, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from tmux pane scrollback and session event logs. It mitigates risks by documenting an explicit evidence chain for safe handling: Ingestion points: Tmux pane buffers and session event streams. Boundary markers: Recommends code fences and structured envelopes for data injection. Capability inventory: High; includes remote build execution and multi-agent orchestration. Sanitization: Provides explicit instructions to avoid shell-interpolation of untrusted output to prevent command injection.
  • [COMMAND_EXECUTION]: Orchestrates numerous CLI tools including git, beads, and tmux to manage multi-agent swarms and project state.
  • [EXTERNAL_DOWNLOADS]: Includes adapters for specialized tools like giil for image fetching and rch for remote compilation as part of its intended integration surface.
  • [DYNAMIC_EXECUTION]: Employs declarative YAML pipelines and prompt templates to assemble and execute complex agent workflows at runtime.
  • [SAFE]: The skill is a well-documented technical reference authored by dicklesworthstone, featuring extensive security mechanisms such as two-person approval gates (SLB), destructive command guards (DCG), and automated credential isolation.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 4, 2026, 09:51 AM