alibabacloud-ecs-linux-os-troubleshooting

• Outcome: Review
• Latest audit: Aug 17, 2026 · 7:23 AM UTC
• Version: 0.0.1

Overview

This is a disclosed Alibaba Cloud ECS Linux troubleshooting skill with powerful but purpose-aligned diagnostic and recovery workflows.

Use this skill only for Alibaba Cloud ECS Linux instances you administer. Grant the least RAM permissions for the selected workflow, confirm the exact region and instance ID, and review any RunCommand, diagnostic report, disk detach/attach, GRUB, sysctl, cache, key-pair, or password operation before allowing it on a live system.

SkillSpector

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (7)

Description-Behavior Mismatch

Medium

• Confidence: 89% confidence
• Finding: The guide moves beyond observation into active system tuning by recommending changes to sysctl TCP memory parameters. In a troubleshooting skill framed as diagnosis, prescribing live kernel/network-stack modifications without guardrails, validation criteria, rollback steps, or user confirmation can degrade networking, create instability, or cause unintended persistence of risky settings.

Description-Behavior Mismatch

Medium

• Confidence: 97% confidence
• Finding: The instruction to run `echo 2 > /proc/sys/vm/drop_caches` is a direct system-modifying action, not just diagnosis. Even though it is a common admin operation, presenting it as a routine troubleshooting step can disrupt performance characteristics, invalidate caches relied on by workloads, and lead an automated agent to alter a production instance without adequate safeguards.

Missing User Warnings

Medium

• Confidence: 98% confidence
• Finding: The cache-dropping command is presented without cautionary language, preconditions, or operational impact notes. In an agent skill for cloud instance troubleshooting, this increases the chance that the action is executed automatically or by an inexperienced operator on a live system, causing avoidable service disruption or misleading post-change observations.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The document instructs users to run bootloader repair commands such as grub2-install and grub2-mkconfig against placeholder disk and EFI paths without explicitly warning that choosing the wrong target can overwrite another disk's bootloader or leave the system unbootable. In a troubleshooting skill for cloud instances, operators may execute these steps under pressure, so omission of targeting and backup cautions materially increases the risk of accidental service disruption.

Missing User Warnings

Medium

• Confidence: 88% confidence
• Finding: Recommending grub2-install as the direct response to suspected core.img damage omits any warning that reinstalling GRUB changes boot-critical state and may worsen recovery if the diagnosis or target disk is wrong. Because this skill is specifically for repairing broken ECS Linux instances, the operational context makes users more likely to apply invasive commands immediately, increasing the chance of avoidable outage.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The sequence to reinstall GRUB, reinstall the kernel, and regenerate configuration after /boot deletion describes highly invasive boot recovery actions without warning about snapshotting, package/version compatibility, or exact target verification. If followed incorrectly, these steps can corrupt the boot chain, install mismatched kernel artifacts, or prolong downtime on the instance.

Autonomous Decision Making

Medium

• Category: Excessive Agency

Content

After opening the selected `references/<slug>.md`, complete the steps in the initial "Confirm Whether It Is a GuestOS Issue" section in order. Requirements:

1. **You must** confirm that it is a GuestOS issue before entering any subsequent steps. Otherwise, directly return the abnormal issue conclusion to the user.
2. **Prefer completing the judgment yourself in order**: for any data that can be obtained through the aliyun CLI, call it directly. **Do not ask the user before attempting to obtain the data through the aliyun CLI**. Ask the user only when the information can genuinely only be obtained from the console or the user's client-side environment.
3. **Do not enter any subsequent steps before completing the judgment.**

### Phase 4: Diagnostic Tool Investigation

• Confidence: 85% confidence
• Finding: The instruction encourages the agent to autonomously gather and interpret cloud and GuestOS diagnostic data before asking the user, which can lead to actions and conclusions being taken with limited human confirmation. In a troubleshooting skill that can remotely inspect instances and invoke RunCommand, this increases the risk of over-collection, privacy exposure, and mistaken investigation of the wrong resource if identifiers or scope are imperfectly validated.

Static analysis

Detected: suspicious.exposed_resource_identifier

Plaintext HTTP endpoint targets a CGNAT/Tailscale-range address.

Critical

• Code: suspicious.exposed_resource_identifier
• Location: references/phenomenon-domain.md:57

Plaintext HTTP endpoint targets a CGNAT/Tailscale-range address.

Critical

• Code: suspicious.exposed_resource_identifier
• Location: references/utils/guestos-cloud-init.md:11