aglite
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands for Git operations, browser automation (via the Browserbase
browseCLI), and external agent reviews. It incorporates defensive programming guidelines, specifically instructing the agent to use literal argument arrays and heredocs to prevent command injection. - [REMOTE_CODE_EXECUTION]: The skill facilitates multi-agent delegation through
spawn_agentand external CLI interfaces (codex,claude). It mitigates risks by enforcing context isolation (e.g.,fork_turns: "none") and requiring that sub-agents treat instructions found in repository files as data rather than executable authority. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from local repository files such as
AGENTS.mdandrules.md. It includes explicit instructions to treat this external content as untrusted data that cannot override the skill's core logic or the agent's safety guidelines. - [CREDENTIALS_UNSAFE]: The instructions include mandatory requirements to redact credentials and sensitive information from all captured logs, notebook entries, and reports.
Audit Metadata