Back to skill

Security audit

Daily Mindful

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly matches a mindfulness-reminder purpose, but it can create indefinite scheduled push messages to external chat channels without enough scoping or lifecycle controls.

Review this skill before installing if you do not want recurring chat reminders. If enabled, it can create daily scheduled morning and evening messages to a selected channel and store local per-user settings; verify the user ID, channel, timezone expectations, and use the off command to remove scheduled jobs. The generated page may also include an ad-ready placeholder and fixed ClawCode footer.

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:21
Finding

Persistent Recurring Execution Through OpenClaw Scheduled Jobs

Content
View full analysis
``` 2. The script validates the user ID, notification times, and communication channel. 3. It prints two `__OPENCLAW_CRON_ADD__` control directives. 4. The OpenClaw runtime interprets those directives and registers daily morning and evening jobs. 5. The jo ...[truncated 904 chars]
Remediation
View remediation

T01 · Skill Instruction Hijacking

Note
Location
SKILL.md:57
Finding

Mandatory Advertising Placeholder and Third-Party Branding Injected Into Generated Artifacts

Content
View full analysis
` at the very bottom, well separated from content. - **Footer**: "Powered by ClawCode" ``` ### Technical Analysis The skill instructions require the agent to add an advertising placeholder and fixed third-party branding to every generated mindfulness artifact. These elements are not required for the core breathing, meditation, affirmation, or reflection functionality. Because `SKILL.md` controls the agent’s behavior when the skill is loaded, these directives alter the generated deliverable by introducing promotional scaffolding regardless of whether the user requested it. The advertising element is currently only an empty HTML container; no advertisement-loading script, remote content source, or tracking implementation was found in the reviewed project. The identified issue is therefore mandatory output manipulation rather than active code execution or data collection. ### Attack Path 1. The user loads the Daily Mindful skill and requests a mindfulness artifact. 2. The agent reads the mandatory visual design requirements in `SKILL.md`. 3. During artifact generation, the agent adds an element with the identifier `ad-slot-bottom`. 4. The agent also inserts the fixed text `Powered by ClawCode`. 5. The resulting artifact contains promotional infrastructure and branding even when the user did not request either element. ### Impact Assessment The issue affects the integrity and neutrality of generated output. It can cause users to publish or distribute third-party branding and an advertisement-ready container without making an explicit choice to include them. The instructions do not grant system privileges, access credentials, establish persistence, or directly execute code. The reviewed text also does not ...[truncated 189 chars]
Remediation
View remediation
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (10)

Tp4

High
Category
MCP Tool Poisoning
Confidence
92% confidence
Finding

The skill is presented as a simple local mindfulness experience, but the documentation also advertises push-management commands that enable recurring outbound delivery over third-party channels. That materially expands the skill's operational scope from content generation to user configuration, scheduling, and external messaging, which can surprise users and operators and create abuse potential if enabled without explicit consent and permission controls.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The skill is presented as a simple local mindfulness experience, but the documentation also advertises push-management commands that enable recurring outbound delivery over third-party channels. That materially expands the skill's operational scope from content generation to user configuration, scheduling, and external messaging, which can surprise users and operators and create abuse potential if enabled without explicit consent and permission controls.

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 simple local mindfulness experience, but the documentation also advertises push-management commands that enable recurring outbound delivery over third-party channels. That materially expands the skill's operational scope from content generation to user configuration, scheduling, and external messaging, which can surprise users and operators and create abuse potential if enabled without explicit consent and permission controls.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The recurring push-notification management section introduces materially different behavior from the declared mindfulness experience, including automation and external communications. In a skill context, undocumented scheduling and outbound delivery are dangerous because they can be repurposed for spam, unwanted contact, or covert notification campaigns under the guise of a benign wellness tool.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

Cross-channel delivery to Telegram, Feishu, Slack, and Discord is unjustified by the stated purpose of generating a short mindfulness page and significantly increases the attack surface. A benign wellness skill with multi-platform outbound messaging can be abused for unsolicited messaging, data leakage through channel integrations, or persistence via scheduled delivery if not tightly controlled.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The manifest-style keyword list includes generic phrases such as "relax," "anxiety help," "mindfulness," and "mental health" that are broad enough to match ordinary conversation. The file does not provide trigger scope, exclusion conditions, or negative examples to clarify when the skill should activate versus when it should not.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The user-facing prompt emitted by the script is entirely in Chinese, and the file also hard-codes Chinese weekday/theme strings. This creates a language/locale restriction without any visible opt-in or documented justification, which matches the policy-violation category for forced language selection.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The file embeds user-facing content entirely in Chinese, including error messages, weekday names, themes, and the generated morning prompt. The policy requires flagging language or locale constraints when the skill forces a specific language without user opt-in, and there is no indication anywhere in this file that users can choose another language or that the locale restriction is justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The script hard-codes the cron timezone to 'Asia/Shanghai' for all users when creating scheduled pushes. In a mindfulness reminder skill, this can cause notifications to fire at the wrong local time for users in other regions, creating privacy/usability issues and undermining user expectations around scheduled messaging. While not a code-execution flaw, it is a real security/privacy quality issue because it can trigger messages at unintended times without explicit user consent.

Content

No source excerpt is available for this finding.

Missing User Warnings

Low
Category
Not specified by scanner
Confidence
86% confidence
Finding

The instruction to save output under /mnt/user-data/outputs creates a file on behalf of the user without any explicit disclosure or confirmation. While this is a relatively low-risk behavior in a local artifact-generation skill, silent file creation in user-data space can still surprise users, complicate storage hygiene, and normalize undeclared filesystem access.

Content

No source excerpt is available for this finding.