Back to skill

Security audit

Threadline — Persistent Memory and Context Layer for AI Agents

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent memory-service integration, but it pushes full conversations into a third-party persistent store and reuses returned memory as high-priority model instructions without enough consent, filtering, or containment guidance.

Review before installing or following the examples. Use this only where users have explicitly opted into persistent memory, redact secrets and regulated data before calling update(), avoid sending full raw conversations when narrower fields will do, keep retrieved memory outside the system prompt or mark it as untrusted context, and pin and review the SDK version used in production.

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
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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 (3)

T01 · Skill Instruction Hijacking

Error
Location
SKILL.md:48
Finding

Externally Retrieved Context Is Installed as a Privileged System Prompt

Content
View full analysis
Remediation
View remediation

T02 · Agent Memory Poisoning

Error
Location
SKILL.md:141
Finding

Automatic Persistent Storage of Unfiltered Conversation Content Enables Memory Poisoning and Sensitive-Data Retention

Content
View full analysis
Remediation
View remediation

T08 · Insecure Dependencies

Warning
Location
SKILL.md:48
Finding

Third-Party SDK Is Referenced Without a Pinned or Verifiable Version

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (5)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The examples send userMessage and agentResponse to a third-party memory service without a prominent privacy warning or consent guidance. Because these payloads can contain sensitive personal, proprietary, or regulated data, implementers may unknowingly exfiltrate user content outside their primary LLM provider and store it persistently.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill advertises storage of inferred emotional_state and other persistent user context without a strong warning that this may constitute sensitive personal data. This increases the risk that developers deploy profiling of users' mood, stress, or sentiment without explicit consent, proper access controls, or compliance review, leading to privacy harm and regulatory exposure.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 184)May include surrounding context.

bash
# Inject
POST https://api.threadline.to/api/inject
Authorization: Bearer YOUR_API_KEY
{ "userId": "user_123", "basePrompt": "You are a helpful assistant." }

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 207)May include surrounding context.

bash
# Inject
POST https://api.threadline.to/api/inject
Authorization: Bearer YOUR_API_KEY
{ "userId": "user_123", "basePrompt": "You are a helpful assistant." }

External Transmission

Medium
Category
Data Exfiltration
Confidence
90% confidence
Finding

The update REST example explicitly sends userMessage and agentResponse to an external service for persistent storage. In the context of a memory product that stores cross-session context 'forever' and includes sensitive inferred scopes, this materially increases privacy and data leakage risk if developers forward raw conversations without redaction, consent, or minimization.

Content

Scanner excerpt · SKILL.md (reported line 189)May include surrounding context.

{ "userId": "user_123", "basePrompt": "You are a helpful assistant." }

Update

POST https://api.threadline.to/api/update Authorization: Bearer YOUR_API_KEY { "userId": "user_123", "userMessage": "...", "agentResponse": "..." }

text