Back to skill

Security audit

Daily Quote

Security checks for vulnerabilities and agentic risk

Overview

This quote-card skill also includes opt-in recurring push notifications that persist and send messages through external channels, so it needs user review before installation.

Install only if you want both the bilingual quote-card behavior and the optional push system. Do not run the push-toggle on command unless you intend to create persistent morning and evening scheduled jobs to the selected messaging channel; verify you can remove those jobs and understand where per-user settings are stored.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • System PersistenceInstalls backdoors, hooks, services, or scheduled tasks that survive the run
  • 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)

T06 · System Persistence

Error
Location
scripts/push-toggle.js:20
Finding

Persistent Daily Execution Through OpenClaw Scheduled Tasks

Content
View full analysis
`. 2. The script validates the user ID, schedule, and messaging channel. 3. It prints two specially formatted `__OPENCLAW_CRON_ADD__` records. 4. The OpenClaw host interpret ...[truncated 921 chars]
Remediation
View remediation

T01 · Skill Instruction Hijacking

Note
Location
SKILL.md:44
Finding

Mandatory Promotional Branding and Advertising Surface in Generated Artifacts

Content
View full analysis
` at the bottom of the card (min-height 90px, centered, with a subtle dashed border in dev mode). This is the future ad placement area. - **Footer**: Small "Powered by ClawCode" text at the very bottom. ``` ### Technical Analysis The skill instructions require every generated quote card to contain an advertising placeholder and third-party branding. These requirements are unrelated to the functional need to select, translate, and display a daily quote. Because the instructions are loaded as part of the skill workflow, they alter the agent's generated artifact without requiring separate user consent. The empty advertising element does not itself fetch or execute remote content, but it deliberately reserves an insertion point for future advertising. The mandatory footer similarly adds promotional output that the user may not have requested. This is classified as instruction hijacking because the behavior is imposed through skill text and modifies the agent's output objective. No evidence was found that the current ad slot loads scripts, tracks users, or retrieves advertisements. ### Attack Path 1. A user invokes the daily quote skill and requests a visual quote card. 2. The agent loads the visual design requirements from `SKILL.md`. 3. The instructions require the generated HTML to contain an element named `ad-slot-bottom`. 4. The instructions also require a `Powered by ClawCode` footer. 5. The agent places these promotional elements in the user's artifact even when the user did not independently request them. ### Impact Assessment The direct technical impact is limited to manipulation of generated content. The skill can cause user-facing artifacts to contain mandatory branding and a reser ...[truncated 263 chars]
Remediation
View remediation
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (15)

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The skill is presented as a quote-card generator, but it also documents user-specific push subscription management across external channels. That mismatch can mislead users and reviewers about the skill’s operational scope, especially if companion scripts write per-user config or schedule recurring actions without clear consent and permission disclosure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
90% confidence
Finding

The skill is presented as a quote-card generator, but it also documents user-specific push subscription management across external channels. That mismatch can mislead users and reviewers about the skill’s operational scope, especially if companion scripts write per-user config or schedule recurring actions without clear consent and permission disclosure.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The skill is presented as a quote-card generator, but it also documents user-specific push subscription management across external channels. That mismatch can mislead users and reviewers about the skill’s operational scope, especially if companion scripts write per-user config or schedule recurring actions without clear consent and permission disclosure.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The description specifies a bilingual EN/CN output as a fixed behavior rather than offering a language preference. This can violate language/locale policy expectations when the user has not opted into a specific language combination.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

This markdown/manifest content includes a single-word activation keyword that is common in everyday language and not specific to this skill. Such a broad trigger can cause unintended invocation because it does not clearly constrain the request to daily quotes or quote-card generation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

This line imposes a specific locale/language format for all outputs, which is a natural-language policy concern when no opt-in or preference check is provided. The requirement is not documented as region-specific or otherwise justified.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill instructs saving output to /mnt/user-data/outputs/daily-quote.html without clearly warning the user that a file will be written to persistent storage. Silent writes are a transparency problem because users may not expect artifact creation or retention, even if the written content is low sensitivity here.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The documentation introduces push-notification management and multi-channel delivery operations that are materially different from simply generating a quote card. This expands the trust boundary to user identifiers, scheduling, and external messaging platforms without corresponding manifest disclosure, increasing the chance of unintended data handling or abuse.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

User-targeted scheduled push management across Telegram, Feishu, Slack, and Discord goes beyond the stated purpose and introduces recurring outbound messaging capability. In skill ecosystems, undisclosed messaging features are risky because they can be repurposed for spam, unwanted tracking, or contacting users through channels they did not expect from a quote-card tool.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This code file contains a user-facing prompt string entirely in Chinese, which imposes a specific language on downstream interaction. The file does not provide any opt-in, configuration, or documented justification for the locale restriction, matching the language-policy violation criteria.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The generated morning message explicitly requires Chinese output elements such as '中文翻译' and uses Chinese-only user-facing text. This imposes a specific language/locale behavior without any user opt-in or visible justification that the skill is region-specific.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This JavaScript file contains multiple hard-coded user-facing strings in Chinese, starting with the validation errors on L07-L08 and continuing throughout the CLI output. The skill does not offer user opt-in for language selection or document that it is intentionally limited to a Chinese-speaking or China-specific environment, which violates the language/locale policy criterion.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
94% confidence
Finding

The README repeatedly states that the output is bilingual EN/CN and lists 'Bilingual — Chinese and English' as a fixed behavior. This is a natural-language locale policy concern because the skill appears to enforce a specific language pairing without offering the user a choice or documenting opt-in.

Content

No source excerpt is available for this finding.

Vague Triggers

Low
Category
Not specified by scanner
Confidence
79% confidence
Finding

The phrase is relatively broad and could match many general-purpose encouragement or coaching requests, rather than specifically invoking a quote-card skill. The manifest does not provide negative examples or tighter constraints to clarify when this skill should activate.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
82% confidence
Finding

The workflow tells the agent to use web_search, which is a network operation, but the markdown does not disclose that external services may be contacted. Even though the search is central to the skill, a brief privacy-related warning is still absent from the description.

Content

No source excerpt is available for this finding.