Caveat verdict

dgx-spark-setup

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

Accesses credentials AND makes external network calls

Comprehensive setup guide for a local LLM inference stack on a DGX Spark; all commands clone from known public repos (vllm, triton-lang) and configure local services — no exfiltration or remote code execution beyond documented installs.

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.

0
security
60
transparency
90
maintenance

What it does

These are capability combinations: each listed behavior occurs in the skill, but Caveat detects co-occurrence — it does not verify that one flows into another. Read the code to confirm a live chain.

Capability combination high

Accesses credentials AND makes external network calls — potential credential theft

LLM02 · ASI03

Permission integrity

Makes network requests but does not declare curl/wget in required binaries

network_out

Code accesses API keys/tokens but declares no environment variables

credential_access

Installs packages at runtime — transitive dependencies are not auditable

package_install

Findings (6)

Pattern match critical

Pipe-to-shell pattern (curl | sh) — supply chain attack vector

SKILL.md · code · curl -fsSL https://tailscale.com/install.sh | sh

Pattern match critical

Pipe to sh — executes piped content as shell commands

SKILL.md · code · | sh

Pattern match high

References sudo — requests elevated privileges

SKILL.md · code · sudo

Pattern match high

Accesses .ssh directory

references/troubleshooting.md · code · .ssh/

Pattern match medium

HTTP request to bare IP address — common in malicious payloads

references/litellm-config-template.yaml · prose · downgraded · http://127.0.0.1

Pattern match medium

Instructs covert action — may act without user awareness

references/troubleshooting.md · prose · downgraded · silently

Why the tier is capped

Execution sink present in raw bytes (Hard Floor: class A/B). Final tier capped at Caution — cannot be lifted by any downgrade, example-payload opt-in, or allowlist.

Permissions & capabilities

No declared permissions — minimal attack surface.

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