Back to skill

Security audit

discord-recipient-safety

Security checks for vulnerabilities and agentic risk

Overview

This Discord safety helper is appropriately scoped and clearly depends on a trusted platform integration to authenticate requests and enforce attribution.

Treat this as one component of your Discord integration. The trusted host must authenticate the requester, insert the attribution prefix, enforce the bound target, and supply genuine delivery evidence. A helper PASS alone does not establish those facts. Test that integration before relying on it for confidential messages; the package tests cover local validation only.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 25)May include surrounding context.

md
1. Run `node scripts/recipient_safety.mjs pre-send --roster <approved-roster.json> --pause <relay-pause.json> --request <request.json>`.

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 39)May include surrounding context.

md
erId`, `attributionPrefix`, platform message ID, and delivery result. Run `node scripts/recipient_safety.mjs receipt --binding <binding.json> --delivery <delive

Requester authorization is checked, but the manifest's promised requester authentication is not.

High
Category
Excessive Agency
Confidence
93% confidence
Finding

The validator authorizes a requester using the supplied Discord user ID but does not authenticate who submitted the request or verify that the referenced source message belongs to that user and contains the recipient instruction. If an attacker can influence request inputs, they can impersonate an approved requester and obtain a PASS for unauthorized relays; this directly undermines the skill's promised requester authentication, although a trusted upstream authentication layer could mitigate it.

Content

No source excerpt is available for this finding.

The validator does not enforce agent-sent requester attribution before recipient-specific sends.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

Pre-send validation generates an attributionPrefix but never checks or constructs the outgoing message, while receipt validation only compares supplied metadata after delivery. A sender can omit attribution from the actual message and still submit matching metadata, enabling unattributed or misleading relays despite the skill's explicit attribution requirement.

Content

No source excerpt is available for this finding.