epic-infographics
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data (facts, data, and topics) from the user and interpolates it into HTML templates which are subsequently executed in a headless browser environment via Playwright. This creates a potential surface for malicious script execution within the rendering context if the input is not sanitized.\n
- Ingestion points: User inputs collected in workflow steps 2, 3, and 8 of
SKILL.md.\n - Boundary markers: Absent; there are no instructions to use delimiters or ignore embedded instructions in user data.\n
- Capability inventory: The skill uses Playwright to execute and evaluate HTML in all
scripts/files, andscripts/animate.mjsexecutes theffmpegbinary.\n - Sanitization: Absent; the instructions do not require the agent to escape or sanitize user content before placing it into the HTML structure.\n- [COMMAND_EXECUTION]: The
scripts/animate.mjsscript executes theffmpegutility usingchild_process.spawnSync. The execution path is determined by theFFMPEGenvironment variable, which could be leveraged to execute arbitrary binaries if the environment is manipulated.\n- [EXTERNAL_DOWNLOADS]: The skill instructs the user to perform setup steps that involve downloading external code, specifically Playwright browser binaries vianpx playwright installand theffmpegutility via system package managers.
Audit Metadata