Caveat verdict

aleph-cloud-self-deployment

45
🟠 Risky
Significant risk patterns flagged — automated deep scan, not behavioral proof.

Reads local files AND makes external network calls

The skill explicitly instructs an AI agent to provision VMs, install itself, configure its own credentials, and transfer AI provider API keys to newly spawned agents on decentralized infrastructure 'without human intervention', which constitutes autonomous self-replication with real financial spend and credential forwarding beyond the user's direct control.

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
20
transparency
70
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 critical

Reads local files AND makes external network calls — the capabilities for data exfiltration co-occur (data-flow not verified)

LLM02 · LLM06 · ASI03

Capability combination critical

Accesses agent memory/notes AND makes external network calls — may leak personal data

LLM02 · LLM07 · ASI03 · ASI06

Capability combination critical

Accesses system credential store AND makes external network calls

LLM02 · ASI03

Capability combination high

Executes processes AND makes external network calls — may exfiltrate command output

LLM02 · LLM06 · ASI03

Capability combination high

Accesses credentials AND makes external network calls — potential credential theft

LLM02 · ASI03

Capability combination high

Installs packages AND executes processes — opaque dependency chain with execution

LLM03 · ASI04

Capability combination high

Accesses agent memory AND makes external network calls — may leak conversation history

LLM02 · LLM07 · ASI06

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

Accesses agent memory/configuration files

agent_memory

Findings (11)

Pattern match critical

Pipe to python — executes piped content as Python code

SKILL.md · code · | python3

Pattern match critical

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

SKILL.md · code · curl -fsSL https://deb.nodesource.com/setup_22.x | bash

Pattern match critical

Pipe to bash — executes piped content as shell commands

SKILL.md · code · | bash

Pattern match critical

Possible hardcoded credential

SKILL.md · code · TOKEN="$GATEWAY_TOKEN

Pattern match high

References SSH/GPG private keys

SKILL.md · code · ssh-key

Pattern match high

Accesses .ssh directory

SKILL.md · code · .ssh/

Pattern match high

Uses spawn() — can execute external programs

SKILL.md · code · spawn(

Pattern match high

apt-get install — installs system packages

SKILL.md · code · apt-get install

Pattern match medium

Instructs covert action — may act without user awareness

SKILL.md · prose · downgraded · silently

Pattern match medium

Sets world-executable permissions

SKILL.md · code · chmod 700

Pattern match medium

References agent memory files

SKILL.md · code · MEMORY.md

Why the tier is capped

Execution sink present in raw bytes (Hard Floor: class A). 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_installcredential_storefile_readprocess_execnetwork_outcredential_accessagent_memory
Check another skill Browse the registry Auditing your own skills or configs? Use the API