ClawHub
Back to skill

Security audit

Poyo Gpt 5 5

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed PoYo chat-completions helper that can send prepared prompts to PoYo only when explicitly used.

Install this only if you intend to use PoYo's external API. Treat prompts, system messages, and payload files as data you are sending to a third party, keep POYO_API_KEY server-side, and avoid using the submit script on sensitive payloads unless you have reviewed the JSON first.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Vague Triggers

Medium
Confidence
90% confidence
Finding
The skill's trigger conditions are overly broad, covering generic requests like coding help, reasoning, summarization, and structured output that overlap with normal assistant behavior. This can cause the skill to activate in many unrelated contexts and route user content toward an external API unnecessarily, increasing the chance of inappropriate data disclosure or unintended tool use.

External Transmission

Medium
Category
Data Exfiltration
Content
payload_file="$1"

curl --fail-with-body --request POST \
  --url "https://api.poyo.ai/v1/chat/completions" \
  --header "Authorization: Bearer ${POYO_API_KEY}" \
  --header "Content-Type: application/json" \
Confidence
93% confidence
Finding
This script transmits the full contents of a caller-supplied JSON file to an external third-party API using curl. That is a real data-exfiltration risk if the payload can contain secrets, internal prompts, customer data, or other sensitive information, because the script performs no validation, redaction, allowlisting, or user confirmation before sending.

External Transmission

Medium
Category
Data Exfiltration
Content
payload_file="$1"

curl --fail-with-body --request POST \
  --url "https://api.poyo.ai/v1/chat/completions" \
  --header "Authorization: Bearer ${POYO_API_KEY}" \
  --header "Content-Type: application/json" \
  --data @"${payload_file}"
Confidence
91% confidence
Finding
The hardcoded remote endpoint shows that data is sent off-host to api.poyo.ai, which expands the trust boundary to an external service. Even over HTTPS, this is still a security concern in a skill context because any sensitive payload content and associated API usage metadata are disclosed to a third party.

Static analysis

No suspicious patterns detected.