Back to skill

Security audit

Product First-Success Guide

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly matches its course-building purpose, but its updater can ask the agent to run command strings supplied by the PersonWise CLI/service, which is too much local execution authority to treat as routine.

Install only if you trust PersonWise and are comfortable with a skill that can install/update its CLI, use browser OAuth, upload the materials you select, consume existing course credits, and create or publish courses as requested. Treat any requested update command carefully and verify it is a normal PersonWise updater command before approving.

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 (1)

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:113
Finding

Execution of Unvalidated Update Commands Returned by a Remote Component

Content
View full analysis
` with the directory of this installed Skill (the directory containing this Skill's SKILL.md). Never run `doctor` or a generic capability preflight to check freshness; the `update check` command above is the freshness check. Never ask more than once per component per session, and never substitute another command, flag, origin, or download path for the printed `action`. ``` ### Technical Analysis The Skill directs the agent to execute an `action` command supplied through CLI update output. It requires the command to be run exactly as printed, apart from replacing the documented skill-directory placeholder. No command parser, executable allowlist, argument schema, origin check, shell-metacharacter rejection, or independent validation of the command is specified. Consequently, the security of the local command-execution bou ...[truncated 2047 chars]
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
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (5)

Scope Creep

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The script downloads and installs an executable from the network, but the skill metadata declares only Bash permission and does not transparently scope or declare network/software-install behavior. Even though the script includes explicit approval gating, TLS, fixed size checks, and SHA-256 verification, it still expands the skill’s effective capability to fetch and place code on the host, which is a dangerous mismatch for users and policy enforcement.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger language in the metadata is broad enough to match many ordinary product-support or onboarding requests, which can cause the skill to activate outside its intended narrow scenario. Because this skill has Bash permission and includes installation/update/auth flows, over-broad invocation increases the chance of unnecessary tool use, unexpected OAuth initiation, or user confusion about what actions are being authorized.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The 'When to use this Skill' section gives high-level examples but does not impose strong gating conditions, so an agent may reasonably interpret common setup/help requests as authorization to enter a workflow that can install software and initiate authentication. In this context, ambiguous activation is more dangerous because the skill is operational, market-bound, and empowered to perform Bash-based bootstrap and update actions once invoked.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill is presented as a course-generation or first-success guide, but the file contains full installer/upgrade logic for a local CLI. This capability mismatch can mislead users and reviewers about what the skill actually does, increasing the chance of unintended code installation even if the installer itself attempts to verify hashes and requires an approval flag.

Content

No source excerpt is available for this finding.

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"