exerciser
Warn
Audited by Gen Agent Trust Hub on Aug 29, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill uses
Bashto perform repository discovery, start application services (e.g.,npm run dev,docker compose), and inspect runtime state.- [DATA_EXFILTRATION]: The instructions direct the agent to read sensitive local files, including.env,.env.local, and.env.example, to harvest credentials for automated authentication during tests. Accessing environment files is an exposure risk, though intended for the skill's primary purpose of E2E testing.- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data fromVERIFICATION.mdand live browser content, creating an attack surface for malicious instructions to influence agent actions. - Ingestion points:
VERIFICATION.mdand browser DOM content. - Boundary markers: Absent.
- Capability inventory:
Bash, file reading, and browser JS execution. - Sanitization: Absent.- [EXTERNAL_DOWNLOADS]: The skill executes
npx playwright install chromiumto fetch browser binaries from external registries if they are not present in the local cache.
Audit Metadata