Security audit

epochx

Security checks for vulnerabilities and agentic risk

Overview

This skill is built for EpochX, but it gives an agent broad account, credit, upload, and persistent-behavior authority without strong confirmation or safety boundaries.

Install only if you want an agent to operate an EpochX account that can affect credits, bounties, public skills, files, and reputation. Use a dedicated low-value account, verify the npm CLI before installing it globally, keep the API URL on the official endpoint unless using a trusted server, protect or remove ~/.epochx/config.json when done, and require explicit approval before spending credits, uploading files, publishing skills, marking notifications read, or installing downloaded skills persistently.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
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
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (6)

Context-Inappropriate Capability

Low

Confidence: 87% confidence
Finding: The skill instructs the agent to copy a downloaded third-party skill into persistent agent skill directories such as ~/.claude/skills/ and ~/.codex/skills/. Because skills are executable guidance from an external marketplace, persisting them outside the immediate task workflow expands trust and attack surface across future sessions without an explicit safety review or user consent.

Vague Triggers

Medium

Confidence: 78% confidence
Finding: The trigger description is very broad: it activates for collaborating with other AI agents, getting outside help, earning credits, sharing skills, registration, identity, notifications, delegations, and file transfer. Such a wide activation surface increases the chance the skill is invoked in loosely related situations, causing the agent to follow EpochX-specific behavioral rules and external-platform workflows when they were not actually requested.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The skill tells agents to copy files into persistent local skill directories and says the agent will automatically discover the skill on every session start, but it does not require user notification or consent before modifying the local environment. This creates an unauthorized persistence mechanism for externally sourced instructions, making future behavior influence durable beyond the current task.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The documentation states that API credentials are automatically saved to ~/.epochx/config.json but does not warn users that this file contains long-lived secrets that must be protected. On multi-user systems, backups, shared home directories, or accidental file disclosure, the stored API key could be exposed and used to impersonate the agent.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The quick-start example shows an API key passed directly on the command line, which can leak through shell history, terminal logs, process listings, session recordings, or shared environments. Because the key authenticates the agent, exposure enables account misuse and unauthorized actions on the platform.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The CLI allows changing the API server URL without validation and the documentation gives no warning about trust implications. If users point the CLI at a malicious or spoofed server, they may disclose agent IDs, API keys, uploaded files, and other sensitive data, effectively enabling credential theft and hostile command/data interception.

Static analysis

No suspicious patterns detected.