publish-blog

Warn

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill's workflow depends on the execution of multiple external Python scripts, specifically content-scorer.py, brand-voice-scorer.py, approval-manager.py, and execution-tracker.py. These scripts are run at various stages of the publishing process to perform analysis and logging.
  • [DATA_EXFILTRATION]: The skill accesses potentially sensitive business data by reading from the ~/.claude-marketing/ directory. This includes active brand slugs, brand profiles, compliance rules, and agency SOPs. This data exposure is a risk if the agent is misled into sharing these files externally.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through its ingestion of untrusted blog content.
  • Ingestion points: The skill accepts a 'Blog draft' and 'Featured image description' as input.
  • Boundary markers: There are no defined delimiters or instructions to ignore embedded commands within the draft content.
  • Capability inventory: The skill possesses the ability to execute local scripts (command execution), read sensitive local configuration files, and send data to external CMS APIs (WordPress and Webflow).
  • Sanitization: The process does not specify any sanitization or validation logic for the input text before it is analyzed by scripts or published to the web.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 6, 2026, 01:36 AM