hackathon-deck-vote

Pass

Audited by Gen Agent Trust Hub on Oct 1, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill installs global NPM packages @antasphere/slideless and @antasphere/hackathon and downloads presentation templates from the Slideless service using the slideless template start command.
  • [COMMAND_EXECUTION]: The skill executes multiple shell commands including npm install, slideless, hackathon, and custom Node.js scripts (fill.mjs, check.mjs).
  • [DYNAMIC_EXECUTION]: The fill.mjs script uses vm.runInNewContext to execute JavaScript code extracted from the index.html file of the downloaded template. Additionally, check.mjs dynamically loads the playwright library using paths computed at runtime.
  • [REMOTE_CODE_EXECUTION]: When processing demo-type decks, fill.mjs executes an external Node.js script (render.mjs) found within the downloaded template directory via execFileSync.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from local repository files (README, routes, dashboard pages) to suggest presentation content. Ingestion points: Project repository files (README, routes, dashboard pages) as defined in SKILL.md (Step 3). Boundary markers: None identified in the instructions for data ingestion. Capability inventory: File system writes (vote.json), network operations (slideless push/share), and command execution. Sanitization: Validation is limited to data types and lengths, without specific protection against prompt injection payloads in the ingested text.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 1, 2026, 04:52 PM