Caveat verdict

auto-arena

88
🟢 Trusted
No high-risk patterns surfaced by the deep scan — automated capability review, not behavioral proof.

An automated LLM evaluation framework comparing models using user-configured endpoints and API keys; package_install is for standard ML evaluation dependencies and all network activity is for the stated benchmarking purpose.

⚠ Flagged for review — coarse, uncorroborated signal, not a confirmed exploit. Review the config yourself before installing.

Automated static analysis — not a human review. Caveat flags capabilities, not confirmed intent, and can produce false positives. Disagree with this verdict? Use Dispute below.

50
security
20
transparency
70
maintenance

Permission integrity

Code accesses API keys/tokens but declares no environment variables

credential_access

Installs packages at runtime — transitive dependencies are not auditable

package_install

Findings (1)

Pattern match critical

Possible hardcoded credential

SKILL.md · code · api_key: "${OPENAI_API_KEY}

Permissions & capabilities

No declared permissions — minimal attack surface.

package_installnetwork_incredential_access

Is this flag fair?

Check another skill Browse the registry Auditing your own skills or configs? Use the API