deer-sense
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses
uv runandnpxto execute commands for auditing UI components and capturing web page state. Examples include runningglimpsefor component isolation audits and UI captures, as well asaxe-corefor automated accessibility scanning. - [EXTERNAL_DOWNLOADS]: The skill recommends installing and running external packages such as
@axe-core/cli,@axe-core/puppeteer, andeslint-plugin-svelte-a11yto support its auditing functions. - [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data by scanning HTML components and capturing console logs from web pages. This creates a surface where malicious instructions could be embedded in accessibility attributes or logs to influence the agent's behavior during the audit process.
- Ingestion points: Scanning local UI components and external URLs using
axeandglimpse capture(SKILL.md). - Boundary markers: The instructions do not specify any delimiters or warnings to treat the tool output as untrusted data.
- Capability inventory: The skill has the ability to execute shell commands via
uvandnpx(SKILL.md). - Sanitization: There is no mention of sanitizing or filtering the audited HTML content or captured console logs before the agent processes them.
Audit Metadata