building-react-quill-canvases

Pass

Audited by Gen Agent Trust Hub on Sep 18, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill defines how to ingest and render data from PostHog via ph.query and ph.loadInsight. While the instructions include security best practices like using Zod for validation and avoiding dangerouslySetInnerHTML, processing external data at runtime creates a surface for indirect prompt injection.\n
  • Ingestion points: Data is ingested through the ph SDK in SKILL.md and references/starter-scaffold.md.\n
  • Boundary markers: Instructions recommend using Zod for data boundaries in references/platform-libraries.md.\n
  • Capability inventory: The skill uses ph.query, ph.state.set, and browser APIs like navigator.clipboard.\n
  • Sanitization: Instructions advise against enabling raw HTML in react-markdown and suggest structural validation.\n- [EXTERNAL_DOWNLOADS]: The skill utilizes several well-known libraries such as Recharts, D3, Three.js, and TanStack, as well as official vendor SDKs (@posthog/canvas-sdk, @posthog/quill). These are standard dependencies for the development platform and are sourced from trusted registries.\n- [COMMAND_EXECUTION]: references/checklist-example.md includes example shell commands (e.g., pnpm install, curl) within a data module. These commands are provided as string literals intended to be displayed in a UI for manual user interaction (via a "Copy" button) and are not executed automatically by the agent or the skill's code.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 18, 2026, 09:48 PM
Security Audit — agent-trust-hub — building-react-quill-canvases