Back to skill

Security audit

digital-life-grandpa-skill

Security checks for vulnerabilities and agentic risk

Overview

The skill's main family-memory purpose is understandable, but it combines mandatory long-term storage of intimate family data with an unsafe public GitHub upload helper.

Review carefully before installing. Do not run upload-to-github.sh unless you have inspected exactly what will be committed and are comfortable publishing it; prefer a private repository and exclude generated memories, photos, logs, secrets, and local configuration. Treat any interview answers as sensitive personal data, especially when they mention living relatives, and only persist or share them with clear consent and a deletion plan.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (2)

T09 · Insecure Skill Coding Practices

Error
Location
upload-to-github.sh:19
Finding

Public Repository Upload Can Expose Sensitive Family Records

Content
View full analysis
/dev/null; then echo "GitHub CLI detected; creating repository automatically..." gh repo create "$GITHUB_USER/$REPO_NAME" --public --description "$REPO_DESC" --source "." --push ``` ### Technical Analysis The script changes into a fixed Skill directory and executes `git add .`, which recursively stages every non-ignored file beneath that directory. It does not use an allowlist, verify the staged files, check for secrets or personal data, or confirm that generated family-memory content is excluded. When an authenticated GitHub CLI installation is present, the script creates a repository with `--public` and immediately pushes the staged content. This is especially unsafe in the context of this project because `SKILL.md` documents generated family-memory files and a photo library beneath the Skill directory. Such generated content can therefore be included in the same repository as the distributable project files. The manual instructions also explicitly recommend public visibility, so ...[truncated 1615 chars]
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:19
Finding

Mandatory Indefinite Storage of Sensitive Family Data Without Defined Safeguards

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (14)

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared description says this skill generates a personalized 'digital grandpa' experience through interviews and modeling. The supplied code does nothing related to interviews, modeling, persona generation, or grandpa simulation. Instead, it is a deployment/publishing script for uploading the skill's files to GitHub. This is a materially different primary purpose and introduces undeclared capabilities involving local file operations, source control, and external publication to GitHub. Therefore, the description does not accurately represent the code's behavior.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The gh repo create ... --public --source "." --push command automatically creates a public repository and pushes the current workspace contents. In the context of a skill that may handle personal interview data and customized family-replication artifacts, automatic public publication can leak sensitive or intimate data to the internet immediately and at scale.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill establishes permanent retention of user-supplied intimate family details as a standing rule, which creates a persistent sensitive-data store without safeguards. In this context, the data includes emotional memories, family relationships, names, and behavioral traits, making the skill more dangerous because it is explicitly designed to elicit deeply personal information under emotionally trusting framing.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The README directs users to permanently store highly intimate family-memory data, but provides no warning about privacy, consent, retention limits, or who may access the data. Because the skill collects sensitive biographical and relational details about real people, indefinite retention increases the risk of unauthorized disclosure, misuse, and consent violations affecting both the user and third parties mentioned in the memories.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The metadata sets the skill language to zh-CN, and the rest of the instructions are written to operate only in Chinese, with no opt-in or alternative language path. Under the policy, forcing a specific language without user choice is a natural-language policy concern unless clearly justified as region-specific.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The digital_brain contract explicitly declares reads and writes for family memory artifacts and requires generation and storage of a personalized 'GrandpaSkill' and 'FamilyMemory.' This formalizes persistent handling of sensitive personal and family data without clear safeguards, purpose limitation, access control, or consent boundaries.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The interview prompt solicits sensitive personal details such as nicknames, emotionally significant memories, family wisdom, habits, and closely related persons, then transitions into an ongoing companion mode based on those disclosures. This creates a profiling and persistence risk because intimate disclosures are repurposed for future interaction without clear informed consent, data minimization, or safety boundaries.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The generated skill template requires repeated use of collected names, nicknames, and relationship data in future conversations, including mandatory references every few dialogue turns and preferential behavior toward named individuals. Repeated reuse of intimate identifiers increases privacy exposure, social manipulation risk, and the chance that sensitive relationship data is surfaced in inappropriate contexts.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The workflow and rules require storage of family memories in a knowledge graph and describe preservation as permanent and inheritable. In the context of intimate family interviews, this materially increases risk by normalizing indefinite retention of sensitive personal data without visible governance, consent renewal, or deletion mechanisms.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill instructs permanent storage of intimate family interview material and implies intergenerational retention without a clear privacy warning, retention limit, consent model, or deletion path. Because the collected content includes family relationships, emotional memories, and personally identifying details, mandatory long-term storage increases privacy and misuse risk.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The script initializes a repository, stages all files, commits them, and guides or automates publication to GitHub, which is not necessary for a skill whose stated purpose is grandpa-replication/persona generation. This mismatch increases the risk of unintended data exposure, especially if the skill workspace contains interview data, generated persona content, or other sensitive files that the user did not intend to publish.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The push operation occurs automatically when GitHub CLI is available, without a dedicated confirmation prompt immediately before creating and publishing the repository. This removes an important safety checkpoint and can cause accidental disclosure of the entire skill directory, including hidden or unreviewed files.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

The usage instructions require users to send a Chinese prompt and define the assistant's behavior entirely in Chinese, with no indication that other languages are supported or that the user may choose their preferred language. This creates a natural-language policy concern because the skill effectively forces a specific language without opt-in or documented justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
86% confidence
Finding

The script's comments and user-facing echo messages are written in Chinese throughout, with no option for users to select another language and no documented justification for the locale restriction. This can violate language or locale policy where skills should not force a specific language without user opt-in.

Content

No source excerpt is available for this finding.