Back to skill

Security audit

auto-updater

Security checks for vulnerabilities and agentic risk

Overview

The skill openly sets up daily auto-updates, but it gives unattended recurring authority to modify SkillHub and every installed skill.

Review before installing. Use this only if you are comfortable with daily unattended changes to SkillHub and all installed skills. Prefer a dry-run or notification-only schedule, require manual approval before applying updates, pin or verify trusted versions where possible, run it with a least-privileged account, and remove the cron job if you do not want persistent automation.

Vulnerability Patterns
  • System PersistenceInstalls backdoors, hooks, services, or scheduled tasks that survive the run
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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
Findings (2)

T06 · System Persistence

Error
Location
SKILL.md:27
Finding

Persistent Daily Update Task Executes Without Fresh User Approval

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 27-35
Vulnerability Type: Persistent scheduled task
Risk Level: High

Vulnerable Code

bash
SkillHub cron add \
  --name "Daily Auto-Update" \
  --cron "0 4 * * *" \
  --tz "America/Los_Angeles" \
  --session isolated \
  --wake now \
  --deliver \
  --message "Run daily auto-updates: check for SkillHub updates and update all skills. Report what was updated."

Technical Analysis

The Skill instructs the user or Agent to create a daily cron task that persists after the initial Skill interaction. The task wakes an isolated Agent session and directs it to update SkillHub and every installed Skill without obtaining fresh approval for the specific changes.

This creates a persistent execution mechanism tied to mutable upstream content. Although the scheduled task is consistent with the stated auto-update purpose and is disclosed to the user, it remains a cross-session persistence mechanism. Its security properties depend on how the scheduled Agent interprets the message and on the trustworthiness of every upstream update available at execution time.

Attack Path

  1. A user asks the Agent to configure the auto-updater.
  2. The Agent runs the documented SkillHub cron add command.
  3. The cron entry persists beyond the current session and wakes an isolated session every day.
  4. An upstream SkillHub or Skill release is compromised or maliciously modified.
  5. The scheduled session retrieves and applies the affected update without presenting its contents for review or requesting fresh approval.
  6. The updated component subsequently operates with the permissions available to the Gateway or scheduled Agent process.

Impact Assessment

The recurring task can cause future software and Skill changes to be applied under the account running SkillHub. The accessible scope can include installed Skill directories, SkillHub configuration, an ...[truncated 415 chars]

Remediation
View remediation

Remediation Suggestions

  • Do not create a recurring scheduled task by default.
  • Require explicit, informed confirmation before installing the cron entry.
  • Prefer a scheduled dry run or update notification that reports available versions without installing them.
  • Require fresh user approval for the exact set of packages and Skills before each update is applied.
  • Run the scheduled check with a dedicated, least-privileged account that cannot modify unrelated files or system-wide packages.
  • Display the complete generated cron configuration, execution identity, update commands, and affected directories before installation.
  • Provide a direct rollback procedure in addition to the documented cron removal command.
  • Record update activity in an auditable log and alert the user when the scheduled task changes or fails.

T08 · Insecure Dependencies

Error
Location
SKILL.md:53
Finding

Unpinned Global and Bulk Updates Trust Mutable Upstream Releases

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 53-68
Vulnerability Type: Unsafe dependency and Skill update process
Risk Level: High

Vulnerable Code

markdown
For **npm/pnpm/bun installs**:

```bash
npm update -g SkillHub@latest

For source installs (git checkout):

bash
SkillHub update

Always run SkillHub doctor after updating to apply migrations.

Skill Updates

bash
SkillHub update --all
text

### Technical Analysis

The global package update resolves the mutable `latest` release, while `SkillHub update --all` replaces all installed Skills with whatever versions the registry currently advertises. The instructions do not require version pinning, cryptographic integrity verification, publisher or provenance validation, release review, staged deployment, or approval of the resolved change set.

Executing these commands through the persistent cron workflow substantially increases supply-chain exposure. A single compromised publisher account, registry entry, package release, source repository, or Skill update could introduce attacker-controlled instructions or executable components after this Skill has already been reviewed.

The file documents a safer `--dry-run` command elsewhere, but the scheduled update message directs updates to be applied rather than requiring the dry-run result to be reviewed.

### Attack Path

1. An attacker compromises an upstream publisher, registry account, source repository, or release pipeline for SkillHub or an installed Skill.
2. The attacker publishes a malicious version and, where applicable, causes it to resolve through the mutable latest release channel.
3. The daily scheduled workflow runs `npm update -g SkillHub@latest`, `SkillHub update`, or `SkillHub update --all`.
4. The update process retrieves and installs the compromised content without a pinned version or mandatory integrity and provenance validation descri
...[truncated 1006 chars]
Remediation
View remediation

Remediation Suggestions

  • Pin SkillHub and each Skill to an explicitly approved version instead of using @latest or indiscriminate bulk updates.
  • Verify package checksums, signatures, publisher identity, and release provenance before installation.
  • Make SkillHub update --all --dry-run the scheduled operation and present its exact change set to the user.
  • Require explicit approval before applying each proposed update.
  • Update one component at a time and validate its behavior before proceeding to the next component.
  • Use a trusted registry with namespace controls and dependency-confusion protections.
  • Run updates in a sandbox or staging environment before promotion to the active installation.
  • Maintain known-good versions and backups so failed or compromised updates can be rolled back.
  • Avoid administrative execution unless a specific update strictly requires it; separate system-wide package maintenance from Skill updates.
Vulnerability Patterns
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (6)

Self-Modification

High
Category
Rogue Agent
Confidence
98% confidence
Finding

The skill is explicitly designed to update SkillHub and all installed skills, which is a self-modifying and environment-modifying capability. Even if intended as maintenance, unattended updates can introduce malicious or broken upstream changes, change agent behavior, and persist through a scheduled cron job, making the context materially risky.

Content

Scanner excerpt · SKILL.md (reported line 7)May include surrounding context.

md
displayName: "auto-updater"
version: "1.0.0"
summary: "每日自动检查更新SkillHub与技能"
description: "每日自动检查更新SkillHub与技能。Automatically update SkillHub and all installed skills once daily。触发关键词: automatically, auto-updater, auto, SkillHub, installed, updater, update,。"
license: "MIT"
tools:
  - Read

Vague Triggers

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger keywords include very broad terms like 'auto', 'installed', 'update', and 'SkillHub', which are likely to match ordinary user conversation and invoke the skill unexpectedly. In this skill, accidental activation is more dangerous because the documented behavior schedules recurring update tasks and performs software modification actions.

Content

No source excerpt is available for this finding.

Self-Modification

High
Category
Rogue Agent
Confidence
97% confidence
Finding

This repeated capability statement confirms that the skill's core function is to automatically update the platform and installed skills, i.e., to modify itself or its operating environment. In context, the danger is elevated because the updates are scheduled to run daily and may occur without real-time human review, amplifying supply-chain and stability risks.

Content

Scanner excerpt · SKILL.md (reported line 185)May include surrounding context.

md
## 核心能力

- Automatically update SkillHub and all installed skills once daily
- 触发关键词: automatically, auto-updater, auto, SkillHub, installed, updater, update,
  skills

Vague Triggers

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The repeated trigger list again uses vague, everyday terms that can spuriously match unrelated requests. Because this skill's actions include scheduling and updating software, broad triggers materially increase the chance of unintended privileged changes.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill describes convenience features but does not prominently warn that it will create a scheduled cron job and automatically modify SkillHub and installed skills. Users may invoke it without understanding that it causes persistent system changes and unattended software updates, which can alter behavior or introduce supply-chain risk over time.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
83% confidence
Finding

The summary and description combine Chinese and English text, and later sections continue this mixed-language presentation without stating that the user can choose their preferred language. This may violate language or locale consistency expectations when no opt-in or justification is provided.

Content

No source excerpt is available for this finding.