consult-design-thinking

Pass

Audited by Gen Agent Trust Hub on Jul 25, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill frequently invokes the Bash tool to execute internal management scripts (e.g., dt-stage-advance.sh, discover-projects.sh, deliverable-graph.py) located in the $CLAUDE_PLUGIN_ROOT/scripts/ directory. These calls are used to maintain the project state and process metadata.
  • [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it processes untrusted data from external sources.
  • Ingestion points: Step 3 (Empathize) involves the intake of consultant-supplied files, pasted text, and URLs as source material.
  • Boundary markers: The workflow mentions research routing rules and a 'bound knowledge base', but it does not specify the use of strict delimiters to isolate raw external content from instructions in the agent's context.
  • Capability inventory: The agent can execute local shell/Python scripts via the Bash tool and modify the local filesystem using Write and Edit tools.
  • Sanitization: No explicit sanitization or filtering of the external input content is defined prior to its processing for synthesis and artifact creation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 25, 2026, 01:51 PM
Security Audit — agent-trust-hub — consult-design-thinking