optimize-web-animations
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands to manage repository changes and verify optimizations.
- Evidence: Execution of
git status,git commit,git read-tree, andrm -f /tmp/<task>-indexfor staging changes. - Evidence: Reliance on repository-defined scripts like
npm run buildandnpm run lintfor verification. - [REMOTE_CODE_EXECUTION]: The skill performs dynamic code execution by injecting JavaScript profiling logic into a browser environment.
- Evidence: Use of
tab.playwright.evaluateto execute complex performance monitoring scripts (e.g.,profileAnimationPage,samplePerformancePage) provided in the skill's reference files. - Evidence: Dynamic module loading via
import()within anode_replcontext to set up the browser client. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted repository data while possessing shell and browser execution capabilities.
- Ingestion points: Processes repository source code,
AGENTS.md, component files, and animation hooks from the local worktree. - Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are present in the processing workflow.
- Capability inventory: Includes shell execution (
git,npm), file system interaction, and browser-based JavaScript evaluation. - Sanitization: No explicit sanitization or validation of the repository-provided content is defined before it influences the agent's decision-making or execution steps.
Audit Metadata