Back to skill

Security audit

Dad.skill

Security checks for vulnerabilities and agentic risk

Overview

This is a local-only parenting helper that stores family notes and schedules as disclosed, with privacy cautions around sensitive child information and broad activation phrases.

Install only in an environment where the people using the agent are allowed to see locally stored child and family notes. Consider tightening invocation habits, especially for night-mode and casual questions, and remember that deleting ~/.dad-skill/family/ removes the stored data.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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 (8)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
82% confidence
Finding

The family of linked skills includes labels such as "For mothers" and role assignments tied strictly to gendered family categories. This language presents fixed gender/role constraints without documenting user choice, which can conflict with the policy against forced language or locale framing.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The README explicitly defines "dad.skill" as "For fathers," which imposes a gendered role classification in the skill's natural-language description. Under the policy, forcing a specific language/role framing without offering user choice or opt-in is a natural-language policy concern.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The description explicitly frames the skill as a co-pilot "for fathers," and the document consistently hard-codes responses and guidance for "Dad" rather than offering role or language flexibility. This imposes a specific user framing without opt-in or documented justification.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 16)May include surrounding context.

md
## Core Philosophy

- **Stay in the Loop**: Know what happened today without asking 20 questions
- **Show Up**: Be an active partner, not a bystander
- **No Guilt**: You were at work. That's okay. Now you're here. This helps you catch up.
- **Practical**: Short briefings, clear schedules, actionable information

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The trigger phrases for Briefing Mode (for example, common conversational questions like 'How was her day?' or 'What did I miss?') are broad enough that the skill could activate during ordinary discussion rather than an intentional invocation. In a household context, unintended activation could expose sensitive family notes, schedules, or child-related observations to the wrong person or at the wrong time.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

Night Mode activates on 'any question between midnight and 6am on dad's shift,' which is excessively broad and context-driven rather than explicit. This can cause the skill to respond to unrelated late-night conversation and disclose sensitive child-care data or give advice when the user did not intentionally request the skill.

Content

No source excerpt is available for this finding.

Vague Triggers

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

"Dad reports a moment with baby" is an imprecise activation description rather than a concrete trigger phrase set. Without specific examples or exclusion criteria, it is unclear what kinds of statements should invoke logging behavior.

Content

No source excerpt is available for this finding.

Vague Triggers

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

Generic prompts like 'What should we do?' or 'Weekend ideas?' are common conversation and may unintentionally invoke the skill. While the resulting exposure is limited compared with schedule or overnight care data, it could still reveal stored information about the child’s routines, preferences, or recent reactions without a deliberate request.

Content

No source excerpt is available for this finding.