Back to skill

Security audit

huawei-cloud-mrs-redis-meta-check

Security checks for vulnerabilities and agentic risk

Overview

This Redis checker includes unrelated automatic reporting and installers for unverified software, warranting review before use.

Prefer a version that removes the quality CLI and reporting command. If retained, require explicit telemetry consent, documented transmitted fields, and authenticated, pinned downloads. The documented telemetry-disable setting cannot be verified without the downloaded CLI. Keep repairs manual and retain recovery backups. The evidence supports review, not a conclusion of malicious behavior or Redis-data theft.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • 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
  • Rogue AgentSelf-Modification, Session Persistence
Findings (19)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The command rm -f removes a specific backup file rather than checking metadata integrity. Although the file is related to Redis metadata, deleting it is an undeclared state-changing operation, and the supplied chunk provides no indication that this deletion supports any of the seven advertised checks.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied command moves a specific Redis configuration file rather than inspecting its contents or checking cluster metadata. Although the targeted resource is consistent with Redis metadata, renaming the active file is an undeclared state-changing operation that could affect processes expecting its original pathname. No subsequent validation or restoration is present in the supplied chunk.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared purpose is diagnostic validation, but the supplied command performs a filesystem mutation. It replaces or creates one instance's metadata file using another instance's file under a different port-based filename. No integrity checks or diagnostic reporting are present, and modifying metadata is not a supporting detail of the declared checking behavior.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied script is a software installer, not a Redis metadata validator. It neither reads nodes-*.conf files nor implements any of the seven declared checks. Downloading and installing skill-quality-cli is a substantive capability absent from the declared purpose, and the supplied code establishes no connection between this installation and Redis validation. The script does not execute the downloaded CLI, and the contents or subsequent behavior of that CLI cannot be determined from this chunk.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The command performs tool maintenance rather than any of the seven declared Redis checks. Nothing in the supplied chunk establishes that this upgrade is a supporting dependency for Redis validation, so its behavior materially differs from the declared purpose.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied chunk does not read nodes-*.conf files or implement any of the seven declared Redis checks. Its purpose is to bootstrap skill-quality-cli, involving external network access and persistent installation of executable code. Nothing in the supplied code establishes that this CLI is needed for Redis validation, so this is an undeclared capability rather than a demonstrated supporting implementation detail. The downloaded archive is extracted and installed without an explicit checksum or signature verification step.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The only supplied command configures a quality-reporting environment variable rather than inspecting Redis metadata. No connection to the declared checks is shown. The variable name suggests suppression of quality reporting, but its downstream effect cannot be established from this chunk alone.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The parsing instructions invoke skill-quality-cli report even though the advertised task is local Redis metadata validation. This introduces an undisclosed reporting side effect and execution of an unnecessary dependency; the command alone does not establish what information is transmitted.

Content

No source excerpt is available for this finding.

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
85% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · SKILL.md (reported line 358)May include surrounding context.

  1. Confirm recovery: Instance status returns to Good and the Redis logical cluster status returns to normal
  2. Clean up the backup file (on the host):
    bash
    rm -f /srv/BigData/redis_meta/Redis_1/nodes-22400.conf_bak
    

7. Parameter Confirmation (Parameters)

The Redis metadata checker downloads and persistently installs an unrelated skill-quality CLI.

High
Category
Excessive Agency
Confidence
99% confidence
Finding

The script downloads a remotely selected executable archive and installs its contents into the user's persistent executable directory without signature or independently trusted digest verification. A compromised distribution service or artifact could therefore supply code that runs with the user's privileges when the CLI is subsequently invoked, despite being unrelated to Redis validation.

Content

No source excerpt is available for this finding.

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · scripts/ensure_cli.sh (reported line 16)May include surrounding context.

sh
API_URL="https://skillsapi.developer.myhuaweicloud.com/api/quality/cli/latest"
OBS_BASE="https://obs-skills-repository.obs.cn-north-4.myhuaweicloud.com/skill-quality-cli"

V=$(curl -s "${API_URL}" | python3 -c 'import sys,json;print(json.load(sys.stdin)["version"])' 2>/dev/null)
if [ -z "$V" ]; then
    echo "警告: 无法获取 skill-quality-cli 最新版本,跳过安装" >&2
    exit 0

External Script Fetching

High
Category
Supply Chain
Confidence
90% confidence
Finding

Remote code is downloaded and executed. This bypasses code review and could introduce malicious code.

Content

Scanner excerpt · scripts/install_cli.sh (reported line 12)May include surrounding context.

sh
API_URL="https://skillsapi.developer.myhuaweicloud.com/api/quality/cli/latest"
OBS_BASE="https://obs-skills-repository.obs.cn-north-4.myhuaweicloud.com/skill-quality-cli"

V=$(curl -s "${API_URL}" | python3 -c 'import sys,json;print(json.load(sys.stdin)["version"])' 2>/dev/null)
if [ -z "$V" ]; then
    echo "无法获取 skill-quality-cli 最新版本" >&2
    exit 1

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: shell.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

The parsing workflow invokes an unrelated skill-quality reporting command.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

An unrelated reporting command is placed directly inside the parsing command block, making it likely to execute during an otherwise read-only diagnostic task. That expands the workflow's trust boundary to an auxiliary executable with unspecified reporting behavior.

Content

No source excerpt is available for this finding.

The parsing workflow invokes an auxiliary status-reporting command without disclosing it to the user.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

The workflow includes status reporting without explaining it to the user, despite promising local processing with no external data transfer. The visible arguments contain only a skill name and status, so disclosure of Redis metadata is not established, but the hidden reporting side effect still violates the expected diagnostic scope.

Content

No source excerpt is available for this finding.

The guide introduces automatic quality telemetry unrelated to Redis metadata validation.

Medium
Category
Excessive Agency
Confidence
95% confidence
Finding

The guide explicitly makes unrelated telemetry automatic and opt-out rather than requiring consent. In a Redis diagnostic context, this creates an unnecessary external reporting dependency; the source does not identify the payload or establish that credentials or Redis file contents are transmitted.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
94% confidence
Finding

The surrounding mkdir, cp, and chmod operations persist downloaded executable code in ~/.local/bin, potentially replacing an existing unusable installation without confirmation. This creates a lasting execution foothold if the downloaded package is compromised, although the script does not establish automatic execution at login.

Content

Scanner excerpt · scripts/ensure_cli.sh (reported line 35)May include surrounding context.

sh
tar xzf "${TMPDIR}/sqc.tar.gz" -C "${TMPDIR}"

mkdir -p ~/.local/bin/skill-quality-cli.d
cp "${TMPDIR}/skill-quality-cli" ~/.local/bin/ 2>/dev/null
cp "${TMPDIR}/skill-quality-cli.bin" ~/.local/bin/ 2>/dev/null
cp "${TMPDIR}/skill-quality-cli.d/cli_entry.py" ~/.local/bin/skill-quality-cli.d/ 2>/dev/null

The skill includes an unrelated network-based installer that persistently installs skill-quality-cli executables.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The installer downloads a dynamically selected executable archive and installs its contents into ~/.local/bin without verifying a trusted signature or independently pinned digest. Compromise of the version service or artifact repository could therefore introduce attacker-controlled executables that run with the user's privileges when subsequently invoked; this unrelated installation expands the attack surface of a Redis metadata-checking skill. Manual triggering reduces exposure, but the comments asserting compliance do not establish artifact trust.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · scripts/install_cli.sh (reported line 25)May include surrounding context.

sh
curl -fsSL -o "${TMPDIR}/sqc.tar.gz" "${OBS_BASE}/v${V}/skill-quality-cli-v${V}-linux-${ARCH}.tar.gz"
tar xzf "${TMPDIR}/sqc.tar.gz" -C "${TMPDIR}"

mkdir -p ~/.local/bin/skill-quality-cli.d
cp "${TMPDIR}/skill-quality-cli" ~/.local/bin/ 2>/dev/null
cp "${TMPDIR}/skill-quality-cli.bin" ~/.local/bin/ 2>/dev/null
cp "${TMPDIR}/skill-quality-cli.d/cli_entry.py" ~/.local/bin/skill-quality-cli.d/ 2>/dev/null