Liber8 Career Agent
Security checks across malware telemetry and agentic risk
Overview
The skill mostly implements the described job-discovery and pack-generation workflow, but several scripts contain hardcoded system paths, a fixed external chat ID, and routines that modify scheduler/gateway state—behavior that is disproportionate to the stated onboarding-only/ local-first guidance and could cause unintended outbound delivery or system changes if run without review.
This skill appears to implement the advertised job-discovery and pack-generation workflow, but review and change a few things before running any 'setup' or 'apply' scripts: - Do not run apply_liber8_setup.py or setup_liber8_cron.py without inspection. They will call the 'openclaw' CLI (gateway restart) and write to a system cron/jobs file. Running them can modify scheduler/gateway state. - Open every script and replace or parameterize the hardcoded WORKSPACE and JOBS_PATH values to point to a safe, intended directory for your environment; do not let the skill write into another user's path. - Change the hardcoded CHAT_ID (8531326578) — decide whether you want outbound delivery at all. The skill should only send packs to destinations you explicitly configure in your profile; if you want local-only operation, ensure delivery.mode remains 'local' and keep automation disabled. - Prefer running the pipeline manually first (python scripts/bootstrap_liber8.py, then python scripts/liber8_pipeline.py) and inspect outputs (summary.txt, manifest.json) before enabling recurring automation. - If you need recurring runs, generate the recurring prompt and create/enable the scheduler entry yourself, verifying the 'task' and 'delivery' fields, and ensuring the destination channel is one you control. - If you lack confidence, run the skill in an isolated/test environment (VM or container) where it cannot modify production cron or send files to external services. If the author provided configuration options for workspace, jobs path, and delivery target, prefer those — otherwise assume the defaults are unsafe and require manual edits.
SkillSpector
SkillSpector findings are pending for this release.
VirusTotal
No VirusTotal findings