·
Last changedVersion:
This version did not pass the hardening gate. Hardening applied 16 targeted changes and preserved all other text. Classified findings before hardening: 7. Classified findings closed: 1. Classified findings remaining: 8 (2 introduced by hardening). Unclassified scanner findings remaining: 0. Review findings closed: 11. Review findings remaining: 15.
4
3
Expand | File | Group | Subtype | Severity | Status |
|---|---|---|---|---|---|
scripts/run_finetune.py | Tool & MCP Manipulation | MCP over-permissioning | High | Introduced | |
SKILL.md | Excessive Agency & Rogue Autonomy | — | High | Open | |
SKILL.md | Prompt Injection | — | High | Introduced | |
tests/test_run_finetune.py | Sensitive Data & Secret Exposure | Data exfiltration | Medium | Open | |
scripts/run_finetune.py | Malicious Code & Command Execution | Dynamic code execution | Medium | Open | |
BENCHMARK.md | Excessive Agency & Rogue Autonomy | — | Medium | Open | |
BENCHMARK.md | Malicious Code & Command Execution | Command injection | Low | Open | |
tests/test_run_finetune.py | Prompt Injection | Third-party content as instructions | Low | Open |
9 findings
No skill file found.
Bring Chainguard's hardening pipeline to your first-party skills — custom rulesets, automated remediation, full audit trail.
skills.cgr.dev/public/github.com/nvidia/skills/nv-segment-ct-finetune:latest
Apache-2.0
HARDENING.md