agent-autonomy-kit-hardened
Pass
Audited by Gen Agent Trust Hub on Apr 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The kit utilizes
openclaw cron addto create scheduled tasks for automated reporting and work triggers. This establishes persistent execution patterns that run unsupervised. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection via the
tasks/QUEUE.mdfile, which the agent reads to determine its next tasks. The skill includes embedded instructions (guardrails) to prevent the agent from executing high-risk commands derived from this untrusted input. - [DATA_EXFILTRATION]: The workflow involves posting progress updates to configured team channels (Discord/Slack), which entails sending project-related data to external services.
- [EXTERNAL_DOWNLOADS]: The documentation references external GitHub repositories for installation and dependency management.
Audit Metadata