Back to skill

Security audit

Partner Training

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly coherent, but its update flow asks the agent to run server-provided update commands without local validation.

Review any PersonWise update prompt carefully before approving it. This skill will use browser OAuth, upload user-selected training materials, consume existing course credits, and may publish or enable link access only when requested; the main risk is approving an opaque update command that changes the CLI or installed Skill.

Vulnerability Patterns
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Findings (1)

T03 · Remote Payload Retrieval and Execution

Error
Location
SKILL.md:104
Finding

Execution of Remotely Supplied Update Commands Without Local Validation

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 returned dynamically by the PersonWise CLI or remote service. Although it requires user approval, it does not require the agent to parse or validate the command before executing it. In particular, there is no local enforcement that the returned command: - Invokes only the bundled bootstrap script or an approved PersonWise executable. - Uses only an allowlisted set of update arg ...[truncated 2476 chars]
Remediation
View remediation
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Scope Creep

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The script performs a network download with curl even though the skill metadata declares only Bash permission. That creates a capability mismatch: invoking this skill can fetch and install remote code, expanding its effective privileges beyond what a reviewer or runtime policy may expect. Although the script includes strong integrity checks, the hidden network/install behavior still increases supply-chain and policy-bypass risk in this skill context.

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"