knowledge-dashboard

Pass

Audited by Gen Agent Trust Hub on Jul 25, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses Bash to orchestrate local Python scripts and utilities for file probing and metadata extraction. It invokes scripts such as render_dashboard.py and build_graph.py from a vendored directory within the plugin root, which is a controlled execution environment. Variables like wiki_path are passed to these scripts, and the skill performs basic file existence checks before execution.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface due to its ingestion of data from research project metadata, binding configurations, and audit reports to generate the knowledge-overlay.md sidecar.
  • Ingestion points: The skill reads from binding.json, project metadata via pipeline-summary.py, and audit logs located in the wiki/audits/ directory.
  • Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions to isolate untrusted content when generating the markdown sidecar.
  • Capability inventory: The agent has Read, Write, and Bash permissions within the knowledge base directory, which are used to generate the dashboard files.
  • Sanitization: The workflow does not explicitly mention escaping or validating project identifiers or audit findings before they are interpolated into the resulting markdown files.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 25, 2026, 01:51 PM
Security Audit — agent-trust-hub — knowledge-dashboard