NYC

x-post-creator-skill

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [NO_CODE] (SAFE): The skill is composed exclusively of text-based reference files and instructional markdown. No Python, Node.js, shell scripts, or binary executables are included.
  • [DATA_EXFILTRATION] (SAFE): There are no network-enabled commands, hardcoded credentials, or references to sensitive system file paths. All operations are local to the provided text context.
  • [PROMPT_INJECTION] (SAFE): Analysis of the instructions reveals no patterns intended to bypass agent safety filters, extract system prompts, or override core safety guidelines. The instructions focus on medical accuracy and tone.
  • [Indirect Prompt Injection] (SAFE): The skill features a feedback loop in references/feedback-log.md which stores untrusted user input. While this creates a theoretical surface for indirect injection, it is considered safe because the skill has zero functional capabilities or tools for an attacker to leverage.
  • Ingestion points: references/feedback-log.md via the Feedback Integration Protocol.
  • Boundary markers: Absent; user feedback is incorporated into future generations without explicit delimiters.
  • Capability inventory: None; the skill has no access to the shell, filesystem (beyond provided files), or network.
  • Sanitization: Absent; the protocol instructs the agent to apply feedback immediately to future outputs.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:17 PM