skills/calicastle/skills/chalk-logic/Gen Agent Trust Hub

chalk-logic

Pass

Audited by Gen Agent Trust Hub on Jul 24, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill is focused on artistic image generation. All instructions pertain to visual composition, material rendering (warm-white mineral chalk), and quality assurance for chalkboard-style graphics. It does not perform network operations, access sensitive files, or attempt privilege escalation.\n- [PROMPT_INJECTION]: The skill ingests untrusted data from user-provided 'briefs' during its first step. While this represents a surface for indirect prompt injection, the risk is mitigated by a structured distillation process (Steps 2-5) that requires the agent to map the input into a rigid 14-axis recipe and a 4-paragraph image generation prompt, preventing the user input from overriding the agent's operational logic.\n
  • Ingestion points: User-supplied brief in Step 1 of SKILL.md.\n
  • Boundary markers: Absent. The brief is processed as natural language without specific delimiters.\n
  • Capability inventory: Image generation, reading local reference files (example-index.md, repair-playbook.md).\n
  • Sanitization: Absent. The skill relies on the structured prompt template in Step 5 to sanitize output by only including information that becomes pixels.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 24, 2026, 10:52 AM
Security Audit — agent-trust-hub — chalk-logic