Back to skill

Security audit

huawei-cloud-msot-msopprof-operator-profiler

Security checks for vulnerabilities and agentic risk

Overview

This profiling skill needs review because its troubleshooting instructions weaken host security and its verification example can erase previous results.

Before use, remove the SELinux-disabling and world-writable-directory workarounds, use a private output directory, and replace automatic deletion with a fresh directory per run. Run only a trusted operator under ordinary user permissions. The referenced analysis script is missing and needs separate review if supplied later. These concrete concerns justify Review; the scanner's critical rating alone does not establish malicious intent.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (2)

T09 · Insecure Skill Coding Practices

Warning
Location
references/troubleshooting.md
Finding

Permission troubleshooting disables host-wide SELinux enforcement

Content
View full analysis

Impact

SELinux policy denials stop blocking operations throughout the host, allowing otherwise-confined processes to perform actions still permitted by discretionary permissions. The instructions provide no restoration step.

T09 · Insecure Skill Coding Practices

Warning
Location
references/troubleshooting.md
Finding

Profiling output directory is made writable by every local user

Content
View full analysis

Impact

Other local users can plant misleading output entries or rename and remove profiling entries regardless of individual file write permissions, compromising profiling-output integrity and availability.

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (6)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

Making the operator executable is a plausible supporting implementation detail. However, setenforce 0 disables SELinux enforcement system-wide when run with sufficient privileges. This security-policy change is materially broader than the declared profiling functionality and is not disclosed in the description. Its conditional troubleshooting framing does not change the command's effect.

Content

No source excerpt is available for this finding.

Profiling troubleshooting recommends disabling SELinux enforcement system-wide.

High
Category
Excessive Agency
Confidence
99% confidence
Finding

The recommended setenforce 0 command puts SELinux into permissive mode system-wide if executed with sufficient privileges, removing mandatory access-control enforcement beyond the profiling process. A routine profiling permission error does not justify weakening isolation for every workload on the host.

Content

No source excerpt is available for this finding.

The troubleshooting instructions disable SELinux enforcement without warning about the system-wide security impact.

High
Category
Output Handling
Confidence
99% confidence
Finding

The text labels disabling SELinux as temporary but provides neither a system-wide impact warning nor a restoration step. An agent or operator following this troubleshooting advice could leave unrelated services without SELinux enforcement after profiling ends.

Content

No source excerpt is available for this finding.

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
98% confidence
Finding

The chmod 777 parameter grants all local users write access to the profiling directory, although profiling normally requires access only for its executing user. This is an unsafe permission workaround rather than evidence of malicious tool use, with risks primarily involving local output tampering and deletion.

Content

Scanner excerpt · references/troubleshooting.md (reported line 143)May include surrounding context.

Check directory permissions

mkdir -p /tmp/profiling chmod 777 /tmp/profiling

text

### Issue: CSV files missing

The instructions make the profiling directory world-writable without warning about unauthorized access or modification.

Medium
Category
Output Handling
Confidence
98% confidence
Finding

Making the predictable /tmp/profiling directory mode 777 permits other local users to list its entries and create, remove, or replace directory entries. This can compromise profiling-output integrity and availability; file-content disclosure additionally depends on individual file permissions.

Content

No source excerpt is available for this finding.

The verification script deletes existing profiling output without warning.

Medium
Category
Output Handling
Confidence
99% confidence
Finding

The verification script recursively deletes output_npu relative to its current working directory without checking whether it contains valuable prior results or requesting approval. Although the target is bounded rather than arbitrary, running this routine verification can irreversibly destroy existing profiling artifacts.

Content

No source excerpt is available for this finding.