Skillv0.1.1

ClawScan security

Ucloud Api Skill · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

ReviewMar 13, 2026, 6:51 AM

Verdict: Review
Confidence: medium
Model: gpt-5-mini
Summary: The skill appears to implement UCloud API calls correctly, but metadata and declared requirements are inconsistent with the runtime instructions and included code — verify credentials and the source before use.
Guidance: This skill will execute bundled Python scripts that call the UCloud SDK and require your UCloud public/private keys (and optionally project/region/zone). Before enabling or providing credentials: - Verify the skill's source and author; the registry metadata currently omits the credential requirements but the SKILL.md and scripts do require keys. - Only provide least-privilege keys (scoped to a test project) and avoid using high-privilege or billing-enabling credentials in production until you've tested behavior. - Expect the agent to be able to make real, billable changes (create/delete EIPs, VMs). Configure the agent to ask for explicit confirmation before any destructive or billable action. - Review UCLOUD_BASE_URL usage: if you or the environment set this variable it can point the SDK to a custom endpoint; ensure it is trusted. - If you need higher assurance, run the scripts in an isolated environment (test project) first and inspect network traffic or run under your own control before granting persistent access. The main issue here is inconsistent metadata (the skill fails to declare required credentials). That mismatch is likely sloppy configuration—not proof of malicious intent—but you should correct/confirm the metadata and follow the precautions above before use.

Review Dimensions

Purpose & Capability: noteName/description match the implementation: scripts call the official ucloud SDK and perform UCloud read/write actions. However, registry metadata claims no required environment variables or credentials while SKILL.md and the code require UCLOUD_PUBLIC_KEY and UCLOUD_PRIVATE_KEY (and optionally UCLOUD_PROJECT_ID/REGION/ZONE/BASE_URL). That metadata mismatch is unexpected and should be corrected.
Instruction Scope: okSKILL.md instructs the agent to use the bundled scripts and the official SDK, prefer read-only operations first, consult official docs for ambiguous fields, and ask the user for missing values. The runtime instructions do not request unrelated files or other service credentials. The scripts only read the declared UCloud environment variables and payload inputs.
Install Mechanism: okThere is no automated install spec (instruction-only for the environment). The skill expects the ucloud-sdk-python3 package to be present or installed with explicit approval — no downloads or arbitrary install URLs are included in the package.
Credentials: noteThe only sensitive environment variables used are UCLOUD_PUBLIC_KEY and UCLOUD_PRIVATE_KEY (plus optional project/region/zone/base URL), which are appropriate for a cloud-control skill. The concern is that the package metadata did not declare these required credentials — this mismatch increases the risk a user will install/enable the skill without realizing it needs live cloud credentials.
Persistence & Privilege: noteThe skill is not forced-always and does not request system-level persistence. It can be invoked autonomously (platform default), which is expected for a cloud-control skill. Because it performs real cloud operations, autonomous invocation increases blast radius: verify agent policies and require explicit user confirmation for mutating (create/delete) actions.