Back to skill

Security audit

Interactive Product Explainer

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly matches its stated PersonWise explainer purpose, but its update workflow gives too much trust to command text returned by the service or CLI.

Install only if you trust PersonWise, are comfortable with a user-level CLI install, browser OAuth, uploading selected product materials to the service, and using a course credit. Review any requested update command especially carefully; a safer design would construct update commands locally from fixed, validated metadata instead of executing command strings returned by the service or CLI.

Vulnerability Patterns
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • 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
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (2)

T03 · Remote Payload Retrieval and Execution

Error
Location
references/workflow.md:26
Finding

Execution of Remotely Supplied Update Commands

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Note
Location
assets/bootstrap.ps1:20
Finding

Insufficient Filesystem Identity Validation in Windows Installer

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • 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
Findings (4)

Sudo/Root Execution

Medium
Category
Privilege Escalation
Confidence
80% confidence
Finding

Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.

Content

Scanner excerpt · assets/bootstrap.sh (reported line 165)May include surrounding context.

sh
echo "Release executable checksum mismatch." >&2
  exit 6
fi
chmod 700 "$candidate"
if [ "$(uname -s)" = "Darwin" ] && [ "$native_signature_status" = "verified" ]; then
  codesign --verify --deep --strict --verbose=2 "$candidate"
  spctl --assess --type execute --verbose=2 "$candidate"

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
91% confidence
Finding

The instruction to complete explicit link/embed/publication requests 'without asking again' can cause the agent to change sharing state autonomously, potentially exposing a generated product explainer more broadly than the user intended at the moment of execution. Because publication and link sharing affect data exposure and access control, skipping a fresh confirmation creates a meaningful risk of unintended disclosure.

Content

Scanner excerpt · references/workflow.md (reported line 138)May include surrounding context.

md
Use presenter choices only for a concrete user requirement. Keep `run wait` active to terminal
state. Default omitted visibility to an explicit `distribution_target` of `private` — an omitted
target resolves to the OAuth grant's publication ceiling, which can be `link`; complete explicit
link/embed/publication requests without asking again. Return the URL matching the final access
mode only when final state proves playability: `access_mode=link` gives the public `share_url`;
`access_mode=private` gives the login-required `editor_url` and must be described as viewable only
after login, and state clearly that outsiders cannot open it. If sharing is needed, enable link access or ask the

Vague Triggers

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

This JSON manifest-style file contains service metadata and an encoded payload, but it does not provide any explicit trigger phrases, activation conditions, or exclusion constraints for when the skill should be invoked. For manifest files, missing specificity about trigger scope can lead to ambiguous or overly broad activation behavior if downstream systems infer invocation from generic metadata alone.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
28% confidence
Finding

The only natural-language policy category in scope is language or locale policy violations. These lines describe an authentication flow but do not impose a language or locale requirement, so this does not clearly violate the stated policy.

Content

No source excerpt is available for this finding.